Nextgov/FCW - All Content

To Ben Lamm, extinction is just an engineering problem

Camille Tuutti — Fri, 17 Jul 2026 18:14:00 -0400

Drill into a piece of ancient bone in one spot and you get nothing. Move a quarter inch over and you get tons. Ben Lamm calls it luck of the draw and that's where resurrecting a species starts. The DNA comes from bone stored in museum drawers, permafrost cores or a cave on the South Island of New Zealand—anywhere something died cold and dry enough to preserve it for a few thousand years.

Lamm is the CEO of Colossal Biosciences, which he started in 2021 with Harvard geneticist George Church to bring back the woolly mammoth. It has since taken on the dodo, the moa, the bluebuck and the Tasmanian tiger. In 2024, it produced the animals that made its name: Romulus, Remus and Khaleesi, three dire wolf pups from a species gone roughly 12,000 years (or gray wolves with 20 edits, depending on which scientist you ask). Scientists pulled DNA from ancient fossils, edited the cells with CRISPR, cloned the embryos and used domestic dogs as surrogates.

None of this is straightforward. Nothing involving gene editing ever is. I reached out to Lamm to get an explainer on how you bring back an extinct species. He's a systems guy, not a scientist. He built and sold five companies before this, the last an AI firm called Hypergiant. Now, Lamm says, the goal is to "do really great work and push societies forward." He runs Colossal like a product company, not an academic lab, which is part of why he frames de-extinction as engineering rather than science fiction.

Lamm has answered the Jurassic Park question untold times, so I don't ask it. I ask what people never think to: what they get wrong.

What they get wrong, he says, is thinking de-extinction and conservation compete. They're the same fight. People land in two camps. One thinks you can bring back anything, which is where Lamm breaks bad news: "Sometimes you make people sad [when] we tell people there is no dino DNA." You can't build a genome out of nothing.

The other camp says the mammoth money should go to species still alive. Same answer either way. De-extinction is one more tool inside conservation. Not an or but an and.

His case: We're in the sixth mass extinction and we could lose up to 50% of biodiversity in the next 25 years. Conservation works but needs new tools. So, Colossal open-sources its tech for partners and runs more conservation projects than de-extinction ones. Right now, 75 partners worldwide use it.

"Most things like mammoths get the headlines," Lamm says. "Saving elephants doesn't get the headlines."

The mammoth is the hard one, though. Ancient DNA shows up degraded and broken, mixed with microbes, defecation and whatever else got into the sample over a few millennia. The first job is filtering, work for Colossal’s top ancient DNA experts.

Then comes the reconstruction, where AI is used. The team sequenced living elephants to build a reference genome, aligned the broken ancient fragments against it and flagged the differences, the places where mammoth DNA diverges from elephant DNA. Lamm calls the output ancestral state reconstruction maps, a best guess at the original genome.

Colossal first targeted around 60 genes. It has now made over 100 edits, some synthesizing a big piece of the genome, others changing a single nucleotide. The traits are the recognizable ones: the shaggy coat, the tusks, the small ears, the domed cranium, the subcutaneous fat layer and the cold-tolerant nerve endings.

Something has to carry the pregnancy. For now, that's an elephant, roughly two years. Colossal is also building artificial wombs, one for birds, one for marsupial-type mammals that gestate outside the body.

None of the first de-extinct animals will come from a machine, though. "Those will be born, hopefully, in the coming generations," Lamm says.

I asked what happens after the calf is born. Romulus, Remus and Khaleesi already show the answer is complicated. The wolves live at a secret U.S. facility on a 2,000-acre preserve, according to a 2025 TIME article, with no plans to release them into the wild. They're a conservation tool, not a species set loose.

Making one animal isn't the point, Lamm says. Herds are, with real genetic diversity, out in the environment and monitored. That means rewilding models, and rewilding models mean people: landowners, indigenous groups, governments, the public, animal rights groups. Colossal doesn't even get to keep its animals.

"We can't be stewards of these animals," Lamm says. The company works with governments to protect and support them. You bring the mammoth back—and then you hand it over.

Rewilding also means arguments about impact. Some models show mammoths lowering Arctic ground temperatures by 9 to 12 degrees, others say less. Strip the big herbivores and carnivores from an ecosystem and it unravels. Put cold-tolerant megafauna back, Lamm says, and the consensus points to a net gain for the Arctic—though nobody has measured how big.

By now, Lamm has put de-extinction on global front pages, built Colossal into a company valued around $10.3 billion and earned a spot on the 2025 TIME 100 Next list. I ask him for his moon shot, the one problem he has yet to solve. He first says artificial wombs. Then he lands on the one he really wants: cryopreservation and reanimation.

“Suspended animation is an area that I'm really, really passionate about,” he says, referring to freezing a living thing and bringing it back to life later. Colossal already does it with cells. What Lamm wants is to do it to a whole animal, freeze it and revive it intact.

"That'd be pretty cool," he says, which is a very Ben Lamm way to describe one of biology's hardest problems. “We don't believe in impossible. Most things are just engineering challenges, right? They're not impossible.”

]]>

IARPA launches new acquisition marketplace

Edward Graham — Fri, 17 Jul 2026 17:31:00 -0400

The Intelligence Advanced Research Projects Activity announced on Friday that it has launched a new platform to help rapidly onboard innovative capabilities from the private sector.

The IARPA Solutions Marketplace is designed to streamline the procurement process by providing vendors with a pathway to showcase their solutions through short video pitches and then receive feedback from relevant agencies.

IARPA, which serves as the Office of the Director of National Intelligence’s research and development unit, also posted a solution notice on SAM.gov. Interested companies can submit their videos through the platform beginning on August 1.

In a statement shared on LinkedIn on Friday, IARPA acting Deputy Director Jack Cooper said “to effectively harness private sector innovation and align it to the Intelligence Community’s needs, we had to create a new, streamlined pathway for industry partners.”

The marketplace is managed by the Applied Research Institute, which said in a press release that “government customers can browse ready-to-award solutions and move toward contract without restarting an acquisition from scratch.”

This is the first time ARI has worked to make its rapid acquisition model available to the intelligence community, although the nonprofit organization has successfully worked over the past few years to launch similar marketplaces with other federal partners.

In an interview with Nextgov/FCW, ARI Senior Vice President of Federal Programs Jason Preisser noted that the organization has launched acquisition platforms for the Pentagon’s Chief Digital and Artificial Intelligence Office, the Air Force, the Defense Advanced Research Projects Agency and the Army Applications Laboratory.

CDAO’s Tradewinds marketplace was the first platform to launch in November 2022, and since then the approach has enabled smaller providers to pitch their service to agencies without first needing to register on SAM.gov to submit a solution.

“Since the inception of Tradewinds, we've seen probably over 500 new entrants decide to submit solutions because of this new simplified model,” Preisser said. “So it's a very open front door.”

He also noted that the overall rapid acquisition model aligns with a November 2025 memo from Defense Secretary Pete Hegseth that calls, in part, for expanding the Defense Industrial Base and making it easier and faster for vendors to receive contracts for innovative products.

Going forward, Preisser said his goal “is to interconnect these marketplaces, so you really do have a one-stop shop,” with the hope of being able to “start to expand out into other civil agencies.”

]]>

Tech bills of the Week: Screen time standards for kids; Teaching elementary school students about AI; Modernizing agriculture and more

Alexandra Kelley — Fri, 17 Jul 2026 14:47:00 -0400

Universal screen time standards

A new bill asks the National Institute of Standards and Technology to develop new voluntary standards that would act as guidance for parents in managing screen time for children.

The Tracking Online Time And Limits Screen Time Act, unveiled on July 15 by Reps. Pat Harrigan, D-N.C., and George Whitesides, D-Calif., asks NIST to solicit feedback from appropriate stakeholders to develop best practices for setting screen time limits on various devices for children, with the goal of developing a universal screen time standard.

Those stakeholders include various government officials — including the chair of the Federal Trade Commission and the surgeon general — as well as privacy advocates, international standards-setting bodies and organizations composed of children's doctors.

“Parents know exactly how frustrating this is,” Harrigan said in a press release. “You can set limits on your child's phone, but then they move to a tablet, a gaming console, or the TV and you're right back where you started. This bill brings everyone to the table to develop one voluntary standard that gives parents a simple, consistent way to manage screen time across devices without sacrificing privacy or creating another government mandate.”

The culmination of NIST’s work with the stakeholders would be a publicly available website detailing the technical standard for screen time.

Elementary school AI education

Sen. Rick Scott, R-Fla., introduced a proposal on July 14 that seeks to amend a longstanding education law to incorporate more details about the dangers and uses of artificial intelligence.

The Safeguard Kids Act would amend the Elementary and Secondary Education Act of 1965 to allow schools to teach students about “dangers, limitations, and responsible use” of AI systems.

“Artificial Intelligence has the potential to be the greatest information innovation since the printing press, but technology is only as good as our ability to use it well and for the right reasons,” Scott said in a press release. “We cannot let AI be the wild west and hope our kids figure it out; that doesn’t work. It’s on us to guide them as they grow. We need to teach kids about the risks associated with AI, so they can be the kind of principled innovators, leaders, and job creators America’s future economy needs.”

In addition to amending the 1965 law, the bill would also permit Student Support and Academic Enrichment Federal Block Grants to be used to fund both AI literacy and specialized counseling programs. Following introduction, the bill was referred to the Senate Committee on Health, Education, Labor and Pensions.

AI for farmers

Rep. Zachary Nunn, R-Iowa, introduced a bill on July 14 that would increase farmers’ access to AI technologies through programs based in the Department of Agriculture.

The Fostering Agricultural Research and Modernization through Artificial Intelligence, or the FARM AI Act, wants to modernize the federal government’s research and workforce programs available to farmers and agricultural workers.

The proposal would formally designate AI as a priority research area within USDA’s Agriculture and Food Research Initiative and expand AI research through the agency’s Agriculture Advanced Research and Development Authority.

It would also outfit the USDA’s Extension System with more resources for farmers to adopt AI into their agricultural practices, expand USDA grants and fellowships with a focus on modernization and AI, and designate a senior USDA official as the inaugural AI in Agriculture Advisor.

“Iowa farmers have always been the first to put new technology to work when it helps them grow more efficiently, conserve resources, and stay competitive,” Nunn said in a press release. “Artificial intelligence is the next frontier. From yield mapping and precision nutrient application to disease detection and water conservation, AI gives farmers better information to make better decisions in real time.”

AI in Medicare Advantage claim denials

A bill introduced by Rep. Herbert Conaway, D-N.J., seeks to amend the Social Security Act of 1935 to incorporate requirements for how AI is used in prior authorization denials by Medicare Advantage organizations.

H.R.9734, introduced on July 16, would target Title 18 of the Social Security Act to codify more rules surrounding AI in Medicare Advantage claim denials.

Following introduction, the bill was referred to the House Committee on Ways and Means.

AI in arbitration

Rep. David Schweikert, R-Ariz., introduced a bill on July 15 that would permit the use of AI in legal arbitration proceedings.

Arbitration is a legal process where two parties agree to have a third party, or arbitrator, evaluate their dispute outside of a court system. Schweikert’s bill would codify the legality of AI’s usage in these proceedings.

Following introduction, the bill was referred to the House Committee on the Judiciary.

Human judgement in AI-enabled lethal force

On July 16, Rep. Don Beyer, D-Va., teamed up with Reps. Tom Barrett, R-Mich., and Sara Jacobs, D-Calif., to introduce the Human Authority over Autonomous Weapons Act.

The bill adds to previous legislation that seeks to regulate how the Pentagon uses AI in offensive systems and procedures, in particular by directing the Secretary of Defense to ensure that AI-enabled military systems and autonomous weapons are all subject to a sufficient level of “meaningful human oversight.”

“No machine should ever be given the power to decide to kill a human being on its own,” Beyer said in a press release. “Allowing AI systems to independently select and engage targets carries serious risks; machines cannot understand morality or the value of human life, nor can they be held accountable for mistakes, war crimes, or catastrophic escalation.”

The bill would specifically amend Title 10 of the U.S. Code, the governing principles of the U.S. Armed Forces, to identify the human commander or operator who is accountable for and in charge of any AI-enabled system. In the bill’s text, it particularly stipulates the need for keeping humans in the loop during the use of lethal force.

“AI mistakes on the battlefield have the potential to be catastrophic: civilian casualties, friendly fire, unintended escalation, and calculations made on faulty data,” said Jacobs in the same release. “These aren’t risks worth taking — not when they can result in unnecessary lives lost or war. This is a common-sense step to help our military keep pace with emerging technologies while upholding our commitments to international humanitarian law and ethics.”

]]>

GSA to take its Emerging Tech Showcase governmentwide

Frank Konkel — Fri, 17 Jul 2026 12:44:00 -0400

As the General Services Administration closes in on achieving its 2026 moonshot goal to save and automate 1 million hours of work for its employees, the agency will showcase and share those internal efforts — and a host of other proven tech efforts — with a governmentwide audience on July 30.

Registration for the Emerging Technology Showcase is open to all federal employees, and though the event will be held in person at the Interior Department’s Yates Auditorium, it will be livestreamed for a virtual audience likely in the tens of thousands.

“The administration has looked at GSA to lead” on approaches that make government more efficient, effective and responsive, GSA Deputy Administrator Mike Lynch said Thursday at the Government Efficiency Summit.

GSA hosted an internal tech demo day in late spring, featuring several agency employees — including Lynch and Chief Information Officer Dave Shive — who showcased several ways technology was being used across the agency to save time, improve efficiencies, deliver better outcomes and services or make employees’ jobs easier. Several thousand employees tuned in, and Lynch said its in the agency’s DNA to share what’s working.

“That's really a core to GSA's mission and a convening function that we've heard: a lot of demand and interest throughout government to expand what we were doing internally and offer it up more broadly,” he added.

One certain talking point for the event will be a step-by-step guide GSA released in June to cut, streamline and automate work called the Elimination, Optimization and Automation playbook. The playbook builds on lessons learned from federal pilots, mature automation programs and GSA’s own extensive internal enterprise efforts to improve operations. One such effort, the agency’s push to save and eliminate 1 million hours of employee work within the calendar year, is pacing far ahead of schedule.

“We're at roughly 850,000 [hours identified], which has been amazing, and we've been blown away with kind of the adoption that's taken place, largely driven by technology,” Lynch said Thursday. “We put that in [the] playbook, and we're now showcasing that as part of the forum on the 30th that should be available to all of government to see where to use that within your own workflows.”

]]>

Trump stretches declassified China intelligence into broader 2020 election claims

David DiMolfetta — Thu, 16 Jul 2026 23:00:00 -0400

President Donald Trump used newly declassified intelligence Thursday night to revive his claims about 2020 election issues, pointing to findings that China possessed or analyzed more than 200 million U.S. voter records. But the documents do not appear to subvert the intelligence community’s prior conclusions that China did not alter votes or interfere with the results of the election.

The declassified records provide substantially more detail about Chinese intelligence collection involving U.S. voter data and reveal internal debate within the intelligence community over how analysts should characterize Beijing’s election-related activities.

But many of the documents reviewed by Nextgov/FCW do not appear to contradict the intelligence community’s longstanding public conclusion that it found no evidence that China altered voting systems, changed ballots or interfered with the mechanics of the 2020 election.

In a speech to the nation, Trump used the disclosure to press Congress to pass the SAVE America Act, casting the records as evidence of broad weaknesses in U.S. election systems and arguing that tighter federal voting requirements are needed ahead of future elections. The SAVE Act would require documentary proof of citizenship to register and photo identification to vote in federal elections, but it faces steep odds in Congress.

“We will be working closely to mitigate any harm, and we’re taking swift action to ensure that sensitive voter data is better protected, so it can never be bought, it can never be hacked, and we can never watch a stolen election again,” Trump said.

The documents suggest Chinese intelligence used U.S. voter-registration information for identity matching and political analysis, while also collecting other large sets of Americans’ personal data. But they do not show that China altered voter rolls, manipulated ballots or changed the election’s outcome.

In April 2020, intelligence agencies concluded that Beijing analyzed several states’ voter registration datasets to conduct public opinion analysis on the 2020 election, but the publicly released intelligence did not establish that China stole the data.

“China has all along adhered to the principle of non-interference in other’s internal affairs. The U.S. election is an internal matter of the U.S. Its outcome is determined by the votes of the American people. China has never and will never interfere in the presidential elections of the U.S.,” Chinese embassy spokesperson Liu Chang said in a statement.

Some of the newly declassified records appear to bolster rather than overturn the intelligence community’s earlier public findings. An August 2020 National Intelligence Council assessment concluded that China preferred Trump’s defeat but “did not intend to try to affect the election,” while separately warning that Russian actors were already amplifying claims that mail-in voting would lead to fraud and that U.S. elections were “rigged” months before Americans cast their ballots.

That same assessment did warn that foreign actors could exploit poorly secured election infrastructure and potentially manipulate vote-counting systems in isolated cases. But it also drew clear limits around that risk: a coordinated effort to change results at scale would be difficult, audits and paper records would likely detect it and attacks on results-reporting systems would probably delay publication rather than affect certified totals.

Trump also said a separate DHS review had identified approximately 278,000 noncitizens registered to vote in federal elections, though it’s not clear how that finding was reached. Federal law already prohibits noncitizens from voting in federal elections, and studies and audits have generally found such voting to be rare.

“Foreign countries have tried to interfere in elections for years. This is not new,” said Rep. Bennie Thompson, D-Miss., the top Democrat on the House Homeland Security Committee. “His repackaging of old lies with old, cherry-picked intelligence to try to confuse the American people will not change the outcome of the 2020 election, will not change the outcome of the nearly 60 lawsuits rejecting his election fraud claims, and will not change the fact that the 2020 election was secure.”

The U.S. intelligence community has previously disagreed over how to characterize China’s actions during the 2020 election. A declassified intelligence assessment released in 2021 concluded that China considered, but ultimately did not undertake, influence efforts intended to change the presidential election’s outcome.

The assessment also found no indication that China or any other foreign actor attempted to alter election infrastructure like vote tabulation machines. Intelligence agencies instead assessed that Beijing continued longstanding efforts to collect information on U.S. voters, public opinion, political parties and candidates while seeking to shape American policy through economic pressure, lobbying and other traditional tools.

Not everyone inside the intelligence community agreed with the assessment’s conclusion on Chinese influence. The National Intelligence Officer for Cyber issued a minority view concluding that China had taken at least some steps to undermine Trump’s reelection prospects, primarily through social media, official statements and state media. The dissent still agreed there was no information suggesting Beijing tried to interfere with election processes.

At the time, then-Director of National Intelligence John Ratcliffe — now Trump’s CIA director — backed that minority view in a memorandum attached to the assessment, arguing the broader intelligence community had not fully or accurately described the scope of China’s activity. Ratcliffe cited an intelligence community ombudsman’s findings that analysts had applied terminology inconsistently, that alternative assessments had faced institutional pressure and that some analysts working on the majority view did not have access to all relevant compartmented reporting.

Several of the declassified records focus on the process behind producing intelligence assessments. Emails declassified Thursday night show officials ultimately agreed to publish a separate alternative analysis, placing the dissenting view on the record without holding up the intelligence community’s broader assessment.

Ratcliffe ultimately concluded that China sought to influence the 2020 election. But neither his memorandum nor the minority assessment concluded that Beijing altered votes, manipulated vote counting or interfered with the technical administration of the election.

“I have long publicly highlighted China’s nefarious efforts to influence the 2020 election against President Trump, as evidenced by my dissent to the flawed January 2021 Intelligence Community Assessment. The documents declassified today shed further light on China’s intentions,” Ratcliffe said in a statement after the speech.

]]>

Pentagon closes cyber apprenticeship applications early after receiving over 15,000

Edward Graham — Thu, 16 Jul 2026 18:39:00 -0400

The Pentagon prematurely closed the job listing for its new Cyber Registered Apprenticeship Program after receiving more than 15,000 applications, according to a Department of Defense official, although the agency is planning to announce additional opportunities in the coming weeks.

The 12-month apprenticeship is overseen by DOD’s Office of the Chief Information Officer, which announced that the application window for the program went live on July 6. Although applications for the first opportunity were slated to be accepted through July 17, the Pentagon ultimately moved up the end date to July 13.

The official said the number of interested applicants “demonstrates there is not only an immense interest in cyber careers but also a real need for the federal government to provide alternate pathways for individuals to learn hands-on cyber skills and join the cyber workforce.”

The Pentagon first announced the launch of the program in April and said the pilot would help fill relevant cyber vacancies across its operations by, in part, emphasizing skill-based hiring. That focus aligns with the Trump administration’s overall effort to prioritize job experience over academic backgrounds in the federal hiring process.

The now closed job posting for the apprenticeship highlighted this lower bar to entry by only requiring that candidates be over 18 years of age, are U.S. citizens and have the ability to obtain and maintain a government security clearance. The program is designed to train apprentices for entry-level DOD positions, including as cyber defense analysts, cyber defense infrastructure support specialists and cyber defense incident responders.

Even before the posting went live, however, the Pentagon reported overwhelming interest in the apprenticeship. Speaking at the SAP NOW summit in Washington, D.C., last month, DOD Chief Information Officer Kirsten Davies said the program had “already generated more than 70,000 inquiries,” even though it had not yet officially launched.

The Pentagon official told Nextgov/FCW on Thursday that opportunities to participate in the program will continue moving forward.

“As one of the nation’s largest employers of cyber talent, the [Pentagon] recognizes the critical need for cyber training and education, especially now that the digital domain is integral to our national defense,” they added. “We are rolling out additional apprenticeship positions over the next few weeks and encourage applicants to apply for open positions, which will be posted on USAJobs.gov.”

]]>

GSA is eyeing OneGov savings beyond software and tech, official says

Ross Wilkers and Edward Graham — Thu, 16 Jul 2026 17:54:00 -0400

The General Services Administration is looking at how it can broaden its OneGov initiative to include additional services from participating companies beyond just their technology offerings, according to a top official with the agency.

GSA launched OneGov in April 2025 to offer federal agencies the opportunity to onboard heavily discounted software and artificial intelligence products from participating tech firms by treating the government as one customer.

Roughly two dozen companies have reached agreements with GSA so far, including Adobe, OpenAI and ServiceNow. Some of these deals include price cuts of as much as 70% to 90%, although these temporary agreements are set to expire at pre-determined end dates.

GSA said in May that the initiative has made AI capabilities available for use to more than 3.4 million federal workers, with the departments of Health and Human Services, Veterans Affairs, Transportation and State among the agencies taking advantage of these offerings. But GSA is also exploring ways of offering additional discounted products to federal agencies through the OneGov agreements.

“What we're really starting to look at now is, ‘How do we expand beyond software?’” GSA Deputy Administrator Mike Lynch said at GovExec’s Government Efficiency Summit on Thursday.

While Lynch did not elaborate on the types of additional services or products that might be acquired through the initiative, he noted that “many of the companies that participate in OneGov offer multiple products to federal government.”

Even though most of the current products and services offered through OneGov deals have specified expiration dates for their price cuts, GSA officials have said they view the temporary agreements as springboards to longer-term contracts. Lynch said the agency has already identified cost savings of an estimated $1.18 billion since the initiative launched last year.

He added that the approach allows for a more coordinated way “to kind of have much larger engagements that sit under the OneGov platform and benefit the federal government,” which he said will hopefully serve as “a further incentive for industry to work with the federal government in terms of the procurement process.”

Lynch also indicated GSA's interest in putting more effort toward marketing the federal government as a customer to industry, saying "we need to tell our story better as a government," because the perception is that agencies are too bureaucratic and burdensome for many companies.

"When I meet with a lot of emerging technology companies that have products that would be a middle-of-the-fairway fit for the government, and you talk about ‘where does doing business with the federal government sit on your growth journey,’ and these businesses that are valued at multiple billions of dollars, and they're like, 'We'll get to it at some point, but we're growing so fast that we're just going,’" Lynch said.

Some companies tell Lynch that "government's going to be too big of a distraction" to make it a priority in their strategy and "that's a real problem for us," he said.

Part of that marketing push involves explaining changes the federal government has made to "reduce friction" and help "make getting started with government easier," Lynch said.

Keeping government information protected remains a clear starting point for any company looking to enter the government market, he added.

"We can't drop the ball on security. It really matters a lot," Lynch said.

]]>

Energy expects $74M in annual savings from new AI and data tools

Christian Robles — Thu, 16 Jul 2026 17:18:00 -0400

The Energy Department estimates that Quanta, its internal enterprise data platform, and Joulix, a generative artificial intelligence suite of tools, will generate $74 million in annual operational savings, an agency official said Thursday.

In implementing internal AI tools more generally, “we weren’t so concerned with ‘oh this is a new technology.’ It was ‘how do we use this to meet the Energy mission,’” Bridget Carper Arnone, deputy chief information officer for architecture, engineering, technology and innovation, said at GovExec’s Government Efficiency Summit.

The department plans to expand Quanta access to all 88 departmental elements by the end of the fiscal year, Carper Arnone said.

Energy partnered with Databricks to create Quanta in 2025 after the Office of Electricity was given three weeks to sift through a billion documents for compliance with an executive order, she explained.

The amount of data the office needed to go through was too large for Joulix to process. Quanta helped the Office of Electricity condense the 200 gigabytes worth of documents to four gigabytes in 12 minutes, she said.

The data processing “allow[s] us to leverage AI after we fix part of the data problem,” Carper Arnone said.

Quanta is currently used in 44 offices across the agency, with more than 500 data sources, and has helped the Energy Dominance Finance Office project how heat waves would affect electric grids, Carper Arnone said. Other DOE offices are eyeing Quanta to improve funding notification processes and correct cybersecurity vulnerabilities that have been exploited, she added.

Energy’s broadening use of Quanta has come as the department increasingly adopts artificial intelligence tools.

Joulix, which includes a generative AI chatbot known as EnerGPT, has 21,000 users across department elements, Carper Arnone said.

While Energy’s national laboratories have access to chatbots other than EnerGPT, employees at those facilities are using the Joulix-based tool to compile emails, draft responses and write one-pagers to their leadership.

Besides EnerGPT, Joulix has been used to help write job descriptions, speed up contract awarding and help offices comply with executive orders and policy directives, Carper Arnone said, adding that the agency receives feedback from employees on new potential AI applications through a dedicated Joulix email.

“We do go, from a governance standpoint, and look at the risk versus the time frame, and then the reward,” Carper Arnone said.

]]>

Cait Conley fears Trump’s cartel focus could create a national security blind spot

David DiMolfetta — Thu, 16 Jul 2026 13:11:00 -0400

Democratic congressional candidate for New York’s 17th district Cait Conley fears the Trump administration’s focus on cartels and fentanyl is coming at the expense of U.S. preparedness for threats from China and Russia — a trade-off she likened to Washington’s overinvestment in counterterrorism after the Sept. 11 terrorist attacks.

“I think we learned this a decade ago, coming out of the counterterrorism fight,” Conley said in an interview discussing how she’d bring her national security policy and military experience to a congressional seat.

A West Point graduate and Army officer who later worked at the National Security Council and the Cybersecurity and Infrastructure Security Agency, Conley is challenging Rep. Mike Lawler, R-N.Y., in the coming general election.

Conley said the federal government spent years rebuilding its ability to compete with major state adversaries after concentrating much of its national security resources on terrorist organizations. She contends that the Trump administration’s focus on cartels and other threats in the Western Hemisphere risks reversing some of that progress.

“I don’t want illegal drugs coming into America, but I also don’t want to cede key national security interests to China and Russia,” Conley said. “And so we have to find a way to do it all. That is the government’s job in keeping America safe.”

“My concern is: what we have seen from this administration is a cozying up to [Vladimir] Putin [and] a willingness to sacrifice national security priorities in order to line pockets of billionaire buddies with China,” she added. “And — in terms of the federal government workforce — a dismantling of the entities that were there to protect this country against foreign malign influence operations, something that we see Russia [and] China use as the tip of the spear in many of their campaigns against America.”

She was referring to last year’s disbandment of the FBI’s Foreign Influence Task Force and the Office of Director of National Intelligence’s drawdown of its Foreign Malign Influence Center. CISA also froze much of its election security work and ended funding for programs supporting state and local election officials.

On Thursday night, President Donald Trump is widely expected to unveil previously classified findings that allies claim would prove the 2020 election was stolen from him, specifically regarding Chinese attempts to meddle in the election. In 2021, the U.S. intelligence community collectively assessed that Beijing “did not interfere with election infrastructure, including vote tabulation or the transmission of election results.” The IC also said that China “did not deploy interference efforts and considered, but did not deploy influence efforts intended to change the outcome” of the election.

Conley expressed concern that political pressure is shaping national security and intelligence assessments. Last year, then-Director of National Intelligence Tulsi Gabbard fired two top National Intelligence Council officials weeks after the group authored an assessment that contradicted Trump’s claims about Venezuelan gang members.

“I do worry about the integrity of the products that are being pushed out,” she said, “not because I worry about the integrity of the public servants doing it, but because we’re seeing a growing political influence on what those things are.”

She pointed to the conflict with Iran and the disruption of commercial traffic through the Strait of Hormuz as an example of a risk that national security planners had anticipated well before the current fighting.

“Any professional in this space who’s looked at Iran, or has been part of these war games or policy discussions … would tell you that this potential outcome was always known,” she said.

Conley said the danger that Iran could disrupt the strait, withstand an extended air campaign and inflict significant casualties had informed U.S. policy discussions for years. She acknowledged that she had no direct knowledge of the administration’s internal deliberations but questioned whether those risks were fully presented to Trump or disregarded.

Tensions in the strait have impacted gas prices across the country, including in New York.

“I don’t know if they failed to tell him the hard truths, or if this administration just didn’t want to hear it,” she said.

Lawler has voted against multiple House war powers resolutions that sought to limit the administration’s ability to continue hostilities without congressional authorization. He has argued that such measures could constrain the U.S. while American forces remain under threat.

Conley’s policy interests extend beyond congressional committees associated with homeland security, intelligence and foreign affairs. If elected, she said she would seek a seat on the House Energy and Commerce Committee.

“I would like to eventually end up on Energy and Commerce,” she said. The panel has jurisdiction over energy, telecommunications, technology and large portions of the nation’s critical infrastructure. Conley said that portfolio increasingly overlaps with national security as cyberattacks, supply-chain vulnerabilities and foreign-made technology create risks beyond the traditional defense establishment.

“National security in the 21st century is so different than the 20th,” she said.

Conley pointed to the country’s reliance on Chinese-made equipment at U.S. ports as one example. Federal officials have spent years examining cybersecurity risks associated with foreign-made cranes and other port systems.

“I don’t believe we need to pay extra to manufacture a basic drinking glass in America,” Conley said, “but I think if it’s necessary to manufacture our critical infrastructure, we should.”

A position on Energy and Commerce would also give Conley a role in debates over artificial intelligence and large data centers, including their effects on electricity rates, water supplies and local infrastructure. It would relatedly help her focus on legislation geared toward utility costs, an area that she says has been a major issue for constituents.

New York Gov. Kathy Hochul recently ordered a temporary pause on certain state environmental permits for new large data centers while her administration develops a regulatory framework.

Conley said policymakers need to assess those facilities’ effects before residents are left to absorb higher utility and infrastructure costs. She framed the issue as an opportunity for Democrats to become “the party of responsible technology” — supporting innovation while establishing standards intended to protect communities.

“Communities are bearing the costs of these decisions without guardrails,” Conley said, “and we’ve got to stop it.”

Her concerns about political influence extend beyond the quality of individual intelligence assessments. Asked how she understands Trump’s tendency to rely on his own instincts in national security matters, Conley’s answer centered less on his psychology than on the officials and institutions around him.

“I also hold a Republican-controlled Congress accountable too for not checking him,” she said, arguing that the GOP “should have reined him in and forced better judgment in execution and decision-making.”

That dynamic, in her view, reflects a broader incentive problem in Washington: elected officials focused on retaining power and senior administration officials unwilling to deliver unwelcome advice. Conley said her work at the White House and CISA gave her a close look at “the people that were there for the right reasons, and the people that were there for the power.”

She tied that argument to her support for congressional term limits and a ban on members of Congress participating in stock trading. Conley has not settled on a specific term-limit structure but said congressional service should not become a lifelong career.

She argues that addressing Trump’s approach requires more than just installing different experts or producing better intelligence; it demands electing officials willing to act, regardless of the political cost.

“At West Point, there’s a cadet prayer, and one of the lines is having the strength to choose the harder right over the easier wrong,” Conley said. “I believe that same ethos is something that members of Congress should always aspire to.”

]]>

From crime labs to space rocks, NIST helps identify mystery chemicals

John Breeden II — Wed, 15 Jul 2026 17:17:00 -0400

On television crime dramas, investigators often recover an unknown powder or residue, send it to a laboratory and learn exactly what it is before the next commercial break. The real science is slower and more complicated, but one part of that familiar plot device is very real.

Forensic scientists can run an unknown substance through a mass spectrometer, generate a distinctive chemical fingerprint and compare the result against an enormous reference library maintained by the National Institute of Standards and Technology.

NIST recently released a major update to that library, adding tens of thousands of compounds that researchers, manufacturers and forensic scientists can use to help identify mystery substances. The expanded collection includes chemical fingerprints associated with everything from synthetic opioids and environmental contaminants to compounds detected on Mars and organic material recovered from an asteroid.

Those fingerprints are formally known as mass spectra. To create one, a mass spectrometer ionizes a chemical compound and breaks it into charged fragments. The instrument then sorts those fragments according to their mass-to-charge ratio, creating a bar-chart-like pattern that can be distinctive to the original substance.

“Just as a person may be identified by comparing their DNA to a database, a chemical compound may be identified by comparing its mass spectrum to the NIST database,” said Bill Wallace, group leader of NIST’s Mass Spectrometry Data Center.

That makes the quality and breadth of the reference library especially important. A laboratory can generate a highly detailed spectrum from an unknown sample, but identifying the substance often depends on having reliable information available for comparison.

The updated NIST Mass Spectral Library, commonly called NIST26, has two major components. Its Electron Ionization Library, which covers compounds that can be easily vaporized, gained roughly 35,000 compounds and now includes more than 382,000. Its Tandem Library, which is used with many nonvolatile compounds that dissolve in liquids, added about 17,000 compounds and now includes more than 68,000 substances and 3.2 million spectra.

The technology has important applications for law enforcement. Federal, state and local crime laboratories increasingly encounter newly developed synthetic drugs that may be unfamiliar to investigators. NIST also works with the Scientific Working Group for the Analysis of Seized Drugs to maintain a freely available reference library of controlled substances, including emerging fentanyl analogues and other synthetic opioids.

The comparison is not necessarily the instantaneous match often shown on television. Scientists still have to consider the quality of a sample, possible mixtures and other evidence before reaching a conclusion. But trusted reference spectra can give forensic chemists an authoritative starting point when they encounter drugs or other substances they have never seen before.

The full NIST26 library is commonly integrated with commercial mass spectrometers and is available through instrument manufacturers and other distributors. But NIST also provides a free public tool that offers a glimpse into how chemical reference data works.

The NIST Chemistry WebBook contains mass spectra for more than 33,000 compounds along with extensive collections of thermochemical, spectroscopic and other scientific data. Users can search by chemical name, formula, molecular weight and several other identifiers.

To test the public tool, I searched for aspirin, one of the most familiar medicines found in household cabinets. The WebBook quickly returned its chemical formula, C9H8O4, along with a molecular weight of 180.1574, its chemical structure and several forms of spectral data.

Selecting the electron-ionization mass spectrum displayed a series of peaks showing the fragments created when aspirin molecules are ionized and broken apart. The pattern bears little resemblance to the tablets found in a medicine cabinet, but it represents a measurable chemical signature that can be compared with data generated from an unknown sample.

Aspirin is hardly a mystery substance. But the same basic process can help scientists identify far more consequential compounds in forensic evidence, environmental samples, drugs, cosmetics, food and body fluids. Some of the newest additions to NIST26 involve substances linked to urgent problems here on Earth. The update includes nitazenes, a potent class of synthetic opioids increasingly associated with fatal overdoses. It also expands the library’s collection of per- and polyfluoroalkyl substances, commonly known as PFAS or forever chemicals.

Other additions arrive with far more mysterious origins.

Among the new chemical fingerprints are thiophenes, ring-shaped molecules containing sulfur that were detected by NASA’s Curiosity rover on Mars. Their presence raises an intriguing possibility because scientists consider them a potential signature of ancient life. They are not proof that life once existed on Mars, but they add another chemical clue to a mystery researchers have been trying to solve for generations.

The library also now includes complex organic compounds found in dust collected from Bennu, a near-Earth asteroid that has traveled through space for billions of years. Known as alkylated polycyclic aromatic hydrocarbons, those compounds may be related to some of the chemistry necessary for life. Scientists are studying whether material carried by asteroids like Bennu could have helped seed the young Earth with some of life’s chemical building blocks.

That means the same kind of reference library that can help identify an unfamiliar substance in a crime laboratory may also help scientists interpret chemical traces recovered from worlds beyond our own.

Maintaining such a wide-ranging collection requires more than simply gathering spectra. NIST scientists evaluate the measurements, examine chemical structures and names, compare related compounds and replace older entries when higher-quality information becomes available. The result is not just a large database, but a carefully maintained scientific reference that laboratories can use with greater confidence.

The mystery substance on a television crime show may be identified in time for the next commercial break. Real science is slower and more careful. But whether researchers are examining evidence from a crime scene, an environmental sample or dust carried home from an ancient asteroid, the search often begins the same way: by comparing the unknown against a trusted record of what humanity has already learned.

John Breeden II is an award-winning journalist and reviewer with over 20 years of experience covering technology. He is the CEO of the Tech Writers Bureau, a group that creates technological thought leadership content for organizations of all sizes. Twitter: @LabGuys

]]>

Trump’s intelligence pick pressed on 2020 election, defends handling of New York Times subpoenas

David DiMolfetta — Wed, 15 Jul 2026 16:05:00 -0400

President Donald Trump’s nominee for director of national intelligence faced pointed pushback from Senate Democrats Wednesday over his answers about the 2020 election results and broader election security concerns ahead of this year’s midterms.

Jay Clayton, who currently serves as U.S. attorney for the Southern District of New York, told lawmakers on the Senate Intelligence Committee that former president Joe Biden was “certified” president in 2020 after the election commenced, but would not outright say Biden won when asked multiple times.

“I’m not an election denier,” Clayton contended in the hearing. But the answer was not satisfactory for multiple senators who got into back-and-forths with Clayton over his responses.

“You refuse to answer a basic question about who won a presidential election, but you ask to lead America’s intelligence community?” said Sen. Jon Ossoff, D-Ga. “Isn’t it humiliating to be unable to answer this question, to have to indulge the president’s delusions? We know, you know, everybody in this room knows the truthful answer to that question. Why can you not give it?”

Election security has become a flashpoint at the Office of the Director of National Intelligence as the agency has grown involved in Trump’s effort to revisit the 2020 election. Former DNI Tulsi Gabbard appeared at a February FBI search of a Georgia election facility and oversaw a review of Puerto Rico voting machines, while Trump more recently authorized acting DNI Bill Pulte to declassify 2020 election records.

The activity has fueled concerns that ODNI, which manages America’s intelligence apparatus that is meant to predominantly track and counter overseas threats, is being drawn further into Trump’s election claims. On Thursday night, the president is expected to unveil what the administration says is evidence of foreign interference in the 2020 election and vulnerabilities in voting machines.

Throughout his testimony, Clayton asserted the job of ODNI is to focus on foreign intelligence threats and not domestic matters.

Days before his June 11 nomination, Clayton spoke on CNBC and appeared to criticize the nation’s election administration processes.

“On the [election] integrity side, we are doing an absolutely terrible job, and the American people are right to question it. How come we can have an audit trail in every other aspect of our lives that’s important?” he said at the time.

Sen. Angus King, I-Maine, questioned why Clayton opined on those matters.

“For the record, I wish you would share any evidence you have that asserts that there was any problem of voter fraud significant enough to change the results of an election in this country,” King said.

Asked by King whether there is a voter fraud problem in the nation, Clayton said, “I don’t think we can say definitively whether there is or is not until we have better processes.”

On cybersecurity, he said that the topic was a top-of-mind issue for him, especially on election influence matters. He said he would commit to assessing whether ODNI’s cyber and foreign influence components — which were downsized under Gabbard — should be resourced again.

In his current position, Clayton oversees one of the nation’s most prominent prosecutorial offices. He previously chaired the Securities and Exchange Commission during Trump’s first term, and he spent more than two decades at law firm Sullivan & Cromwell, focusing on corporate transactions and capital markets.

Clayton argued that he worked closely with intelligence and law enforcement on terrorism, espionage, cyber threats and illicit finance cases during that work history. He has no prior experience working inside the intelligence community. Clayton’s office is notably prosecuting deposed Venezuelan leader Nicolás Maduro and his wife on narcoterrorism, drug-trafficking and weapons charges following their capture by U.S. forces in January.

If confirmed, he would inherit an intelligence office already undergoing significant workforce changes. Pulte, a major Trump ally who led the administration’s controversial mortgage fraud investigations last year, has initiated multiple rounds of personnel cuts and received broad permission from Trump to declassify records, prompting concerns from former officials about the exposure of sensitive intelligence capabilities.

Clayton was also questioned during the hearing about his office issuing subpoenas last Friday seeking grand jury testimony from four New York Times journalists who reported that Trump returned from a NATO summit aboard an older Air Force One after security officials raised concerns about the Qatari-gifted aircraft that was meant to replace it. The Times reported that the newer plane lacked some of the prior aircraft’s advanced security features, including antimissile capabilities. Some subpoenas were delivered to the journalists’ homes.

Sen. Ron Wyden, D-Ore., criticized the subpoenas as an attack on press freedom and the First Amendment.

Sen. Kirsten Gillibrand, D-N.Y., also questioned how quickly they were issued, arguing that Clayton’s explanation was difficult to reconcile with the fact that security concerns about the Qatari-donated plane had been known for months.

Clayton defended his handling of the matter, saying he was “comfortable” with the process. He declined to discuss specific details because the investigation into the disclosure of classified information remains ongoing, but said he consulted other lawyers before issuing the subpoenas and respects the First Amendment and the press’s role.

“I’m not going to get into the details. But what I can tell you is that we followed the procedures,” he said.

Trump nominated Clayton last month after his decision to install Pulte as acting intelligence chief drew resistance from Democrats and some Republicans. Clayton’s first scheduled hearing was abruptly postponed after Trump said the Senate should first approve his chosen replacement at the U.S. Attorney’s Office in Manhattan and demanded action on unrelated election legislation.

Sen. Mark Warner, D-Va., the vice chairman of the intelligence panel, told reporters after the hearing that he was disappointed in Clayton’s responses and has “a lot to think about.” The senator previously noted that he has known Clayton for many years and believes “he is a capable public servant.”

“I was very disappointed with a number of Mr. Clayton’s answers. It’s not in many ways reflective of the person that I’ve worked with over the years,” Warner said. “I worry that there seems to be this criteria that this administration is putting out, about no one who’s put up for any position can tell the truth about the 2020 election. That’s deeply unsettling.”

]]>

Feds tie counterterrorism grants to election security measures

Chris Teale — Wed, 15 Jul 2026 16:00:00 -0400

The Federal Emergency Management Agency announced late last week that states must implement various election security measures before they can receive homeland security grants.

In its funding opportunity for the $1 billion Homeland Security Grant Program, FEMA and the Department of Homeland Security, its parent agency, said states would need to take what FEMA called “critical, common-sense steps to protect U.S. elections,” to receive any funds.

Those steps include submitting a plan for transitioning away from electronic voting systems that use bar codes and QR codes to count votes and onto equipment that uses hand-marked paper ballots. They also would be required to manually audit at least 5% of all ballots cast after every federal election and reconcile the number of voters who participated in each federal election with the number of ballots cast.

Within 120 days of accepting a grant award, states would be required to use the U.S. Citizenship and Immigration Services’ Systematic Alien Verification for Entitlements system to verify the citizenship status of everyone listed in the state voter registration database. States would also be required to use the SAVE system or another system to verify the citizenship of anyone working at polling places or operating election systems.

“Election security is national security and protecting the Nation’s critical infrastructure is a top priority,” Homeland Security Secretary Markwayne Mullin said in a statement. “Under President Trump’s leadership, we are taking decisive action to protect election systems from threats like foreign interference, insider threats, and cyberattacks. These new requirements for homeland security grant recipients will preserve election integrity and ensure that Americans can trust the results.”

Government groups have already raised objections and concerns that these new requirements might divert a public safety program away from its original mission.

“Election administration has long been a state and local responsibility,” said Irma Esparza Diggs, director of federal advocacy at the National League of Cities, in an email. “Using homeland security grants to influence state election policies raises concerns about federal overreach and the potential erosion of state and local authority. Homeland security funding should remain dedicated to its core purpose of protecting communities from terrorism, cyber threats and other public safety risks.”

This latest gambit comes after a tumultuous week in election administration, after President Donald Trump fired two members of the Election Assistance Commission and another resigned, leaving the body without any sitting commissioners just months before November’s midterm elections.

In addition, the Trump administration has taken legal action to try to obtain voter data from secretaries of state, though it has so far been unsuccessful in court against states opposed to those measures. The administration also cut staff at the Cybersecurity and Infrastructure Security Agency and ended federal funding for the Elections Infrastructure Information Sharing and Analysis Center.

This move to tie grant funding to election-related demands has echoes of an executive order Trump signed last year. That order mandated changes to the national registration form and decertification of vote tabulators that use QR codes. When the EAC did not act on those mandates, the Trump administration earlier this year ordered the U.S. Postal Service to require states to provide citizenship lists for residents voting by mail. Parts of that order were also blocked in court.

Using these homeland security grants to change states’ elections policies is a new move, however. The grants are typically used by state, local, Tribal and territorial governments to prevent, protect themselves against, mitigate, respond to and recover from acts of terrorism. It includes three programs designed to fund various activities for preparedness, including planning, organization, equipment purchase, training, exercises and management and administration.

“The Trump administration says these grant conditions take ‘decisive action’ to protect elections from foreign interference and cyberattacks, but they do the opposite, interfering with national security at the moment officials can least afford it,” said Tim Harper, project lead for elections and democracy at the nonprofit Center for Democracy and Technology, in an email. “[Now], this notice threatens to withhold 20% of a state’s entire homeland security grant, taking hostage the very funds that support fusion centers and other unrelated national security funding for things like bomb squad equipment and active-shooter preparedness. If this administration actually wants to strengthen election security, it would restore the partnerships and cyber funding officials depend on, not threaten public safety in the name of advancing partisan election rules.”

Other local officials said they are reviewing the new guidance from FEMA but are concerned with some of the parameters.

“Counties are on the front lines of both election administration and homeland security preparedness, and many rely on Homeland Security Grant Program funding to protect critical infrastructure, train first responders and secure public spaces,” Eryn Hurley, chief government affairs officer at the National Association of Counties, said in an email. “[Our] priority is ensuring that any new compliance requirements come with adequate guidance, reasonable implementation timeframes and support so that counties aren’t forced to choose between meeting new federal mandates and maintaining the public safety services these grants are designed to fund.”

]]>

GSA’s draft AI procurement rule has improved but needs further reforms, contractors say

Christian Robles — Wed, 15 Jul 2026 15:12:00 -0400

The General Services Administration’s proposed large language model-specific acquisition regulations unveiled last month are an improvement over rules introduced in March, contractors and experts told federal regulators during a Tuesday listening session.

But those stakeholders also told GSA officials that the draft rule requires further modifications that clarify technical definitions and align with commercial best practices.

“Stakeholders asked for clarity, practicality and alignment with commercial norms, and this updated draft shows meaningful movement in all of those directions,” Amy Benson, government affairs vice president for Science Applications International Corporation, said at the listening session.

GSA published a draft “Basic Safeguarding of Artificial Intelligence Systems” rule in March before releasing a revised version titled the “Basic Safeguarding of Data within Large Language Model Artificial Intelligence Systems (LLMs)” last month.

If finalized, the proposed rule would require contractors to follow data protection and intellectual property safeguards when they use large language models to process government data. For example, the draft stipulates that contractors cannot use government data to train or fine-tune large language models or inform business decisions.

The proposal applies to GSA’s government-wide contracts such as the Federal Supply Schedule, government-wide acquisition contracts and the OASIS+ professional services vehicle.

The latest GSA draft softened language around contractors’ use of foreign AI components, assigned flowdown requirements based on four general roles within the industry — LLM Developers, LLM System Operators, LLM System Integrators and LLM Service Providers — and axed a specific prohibition of “diversity, equity and inclusion” ideology in AI models in favor of more generally prohibiting “ideological dogmas.”

Contractors said they welcome many of those changes but want GSA to clarify definitions and reform flowdown requirements.

“We would argue that the definition of data, although improved, is still overly broad and must be tailored to align with commercial practices,” Megan Petersen, a senior vice president at the Information Technology Industry Council, said at the listening session.

She added that the definition of data outputs — which the draft sets as “all data, information, [personally identifiable information], any improvements, enhancements, corrections, annotations, or other modifications made to Data Inputs, or content generated by the LLM in the performance of this contract” — is unclear and could create confusion as to when information becomes government data that requires additional safeguards. Petersen urged GSA to ensure the definition of data outputs excludes metadata, logs and other types of information.

Tim LeMaster, vice president of federal engineering at Lookout, similarly told Nextgov/FCW that GSA should clarify what "incidental" large language model functionality exempt from the proposed rules means in practice.

Benson said the proposal’s definition of custom development — which refers to “modifications, customizations, configurations, or enhancements” made in service of the government contract — is too broad by continuing to cover reusable integrator-developed intellectual property. Another concern Benson raised is that unbiased AI principles language and testing requirements are unclear.

“We just totally nailed the unbiased AI principles,” Jeff Koses, a senior GSA procurement executive, jokingly told listening session attendees.

“[I] heard that we need to continue to work on definitions and recognize what falls outside of those definitions… [I] heard quite a bit of inconsistency with various commercial terms, we need to work up to data rights, including metadata,” he added.

Contractors have until Aug. 3 to comment on the draft rule. GSA has only received 16 comments as of Wednesday afternoon, but several industry groups told Nextgov/FCW they plan to submit comments in the coming weeks.

As GSA mulls industry feedback, experts are urging the agency to offer examples of how the large language model proposal would work in practice.

“Adding examples for what's in and what's out, I think that that's generally a good practice, especially with new regulation,” David Timm, a partner at Burr & Furman, said in an interview.

But regulators should also consider a three-pronged test to determine what data requires additional protections under the proposal, Jessica Tillipman, associate dean for government procurement law for George Washington University law school, said.

]]>

VA wants more input on proposals to streamline benefits application forms

Edward Graham — Wed, 15 Jul 2026 15:07:00 -0400

The Department of Veterans Affairs is pushing for more public feedback on its proposals to reduce the amount of paperwork that veterans and their survivors must complete to apply for benefits.

VA published two recommendations last month proposing to shorten the length of the forms needed to apply for disability compensation benefits and for Dependency and Indemnity Compensation, survivor’s pension and accrued benefits. The department highlighted these ongoing efforts on Wednesday and noted it is accepting additional public input on its recommendations.

VA’s proposal to reduce the amount of paperwork needed to complete the disability compensation benefits form was published in the Federal Register on June 23, with the department saying its Veterans Benefits Administration component is looking to reduce the complexity of the paperwork by the end of calendar year 2026. The outlined recommendation would shorten the form from 15 pages to five pages and decrease the average amount of time needed to fill out the paperwork from 25 minutes to 15 minutes.

“The primary objectives are to enhance the experience for Veterans, lessen the administrative burden and time required to submit a claim for benefits, and accommodate preferences for submitting forms either through VA.gov or via paper, while preserving accessibility,” according to the notice.

The proposal to shorten the DIC application was similarly published in the Federal Register on June 25 and also notes VBA’s desire to reduce the complexity of the form’s paperwork by the end of this year. The notice said the department is looking to shorten the application from 20 pages to seven pages and reduce the average response time from 40 minutes to 25 minutes.

VA said in both proposals that it is looking for comments on how to revise its information gathering process, such as “ways to minimize the burden of the collection of information on respondents, including through the use of automated collection techniques or the use of other forms of information technology.”

Public input on the recommendations, however, has been relatively muted thus far. As of this article’s publication, the two proposals had a combined nine comments — two for the disability compensation benefits recommendation and seven for the proposed DIC application changes. Feedback on both proposals will be accepted through Aug. 24.

VA has already been working over the past several years to shorten or digitize many of the forms that veterans and their beneficiaries must complete to apply for and receive services. These efforts have included a push to create standardized claim and appeal forms, as well as an effort to pre-fill some forms with information from verified VA.gov profiles.

“No Veteran or surviving family member should face unnecessary bureaucratic hurdles when seeking the benefits they’ve earned,” VA Secretary Doug Collins said in a statement. “These proposals reflect our commitment to putting Veterans and their families first — reducing paperwork, saving time, and making the application process as straightforward as possible.”

]]>

White House announces ‘Gold Eagle’ AI clearinghouse for cyber vulnerabilities

Alexandra Kelley and David DiMolfetta — Tue, 14 Jul 2026 18:02:00 -0400

The White House on Tuesday launched an AI-backed clearinghouse designed to consolidate software vulnerability findings from government and industry, prioritize the most consequential flaws and coordinate remediation across critical infrastructure.

The initiative, dubbed Gold Eagle, has already begun processing vulnerability reports, the administration said.

Gold Eagle was established under a June 2 executive order that, in part, directed the federal government to promote the development and secure use of advanced AI systems. The initiative brings together the White House, Cybersecurity and Infrastructure Security Agency and the departments of Treasury and Defense with unnamed open-source software organizations and critical infrastructure providers.

The White House press release announcing the initiative says that those partners had developed a system to “receive and patch” vulnerabilities, though it largely describes Gold Eagle as a coordination mechanism. It does not indicate that the initiative would compel companies to address vulnerabilities directly.

The administration also did not specify which agency would oversee its daily operations, how sensitive vulnerability information would be protected or how the initiative would interact with CISA’s existing vulnerability-disclosure and remediation programs. Gold Eagle joins several existing federal vulnerability programs, including CISA’s disclosure program and exploited-vulnerability catalog, the CVE system and NIST’s National Vulnerability Database.

The announcement did not disclose how many findings Gold Eagle has processed, which companies are participating or whether any vulnerabilities have resulted in completed patches.

But Anthropic is likely to be among the initiative’s private-sector participants, given its past commitments to support vulnerability disclosure efforts. The AI company said last month that it would provide federal officials with advance access to its threat-intelligence reports and participate in the interagency vulnerability clearinghouse created by Trump’s June executive order.

“When significant jailbreaks or misuse patterns are identified, we will quickly investigate, triage, and notify appropriate government counterparts,” the company wrote in a June 30 blog post after a recent export control spat with the White House. “We will also provide government partners with our threat intelligence reporting in advance of publication and participate in the interagency cybersecurity vulnerability clearinghouse established under Sec. 2(d) of the June 2 Executive Order.”

Anthropic did not respond to a request for comment by the time of publication.

Increased coordination between industry and government to find and fix vulnerabilities with the help of AI tools follows the spring debut of Anthropic’s Mythos, a powerful cyber-focused AI model. Initially only available to select private sector partners via Anthropic’s Project Glasswing, multiple companies and some federal agencies have been granted access to Mythos for various testing efforts. Releases of cyber-AI models from other major providers have also followed.

Agencies have also recently been placed under more intense pressure to close cyber openings quickly. CISA recently revamped its remediation timeline guidance, ranging from three days for the highest-risk flaws to 60 days for lower-priority issues.

]]>

Modernization redefined: Why public sector IT leaders must put data at the center of AI architecture

Riccardo Di Blasio — Tue, 14 Jul 2026 17:05:00 -0400

Artificial intelligence is changing how public sector organizations think about enterprise architecture. Public sector IT leaders used to face a straightforward decision: Should this workload live on-premises for control, in the public cloud for scale and speed, or in some hybrid mix?

That decision framework is now obsolete. For AI, success now depends less on where workloads run and more on whether data can support operations across mixed environments without creating extra problems.

At the same time, the conditions shaping infrastructure decisions have grown volatile. Pricing models change without warning, and capacity limits appear suddenly. The rapid build-out of AI infrastructure is driving demand faster than agencies can plan. These pressures are forcing public sector organizations to rethink their entire approach—with data adaptability plus strong governance at the core.

From placement to adaptability

Traditional enterprise architectures were built around separate applications and data silos. Data was locked inside specific workflows. AI breaks this old model. Large-scale AI projects need data to work smoothly across distinct platforms and locations, and many organizations still struggle with doing so.

What looked flexible on paper often turns rigid in practice. Workloads that at one time seemed well placed become expensive or hard to move. Growing data volumes and hybrid cloud make the problem worse. In fast-changing conditions, poor placement decisions quickly become real risks.

Modernization must now shift from placing workloads to building true adaptability. Portability lets you move things, but adaptability keeps systems running reliably during those moves. Organizations need the ability to adjust quickly and access data across environments while ensuring it is stable, governed and secure.

High-performance systems, especially for AI training and inference, often trade flexibility for speed. Some lock-in is intentional, but organizations must balance performance with adaptability. The real limit in most AI setups is not compute power, which can typically be added when needed. Now it is whether data can be accessed, governed and reused across systems without major extra effort each time.

Redefining modernization around data

Public sector organizations must put data first when making infrastructure choices. Plan your exit as carefully as your entry. Make sure workloads and their data can move later without huge cost or downtime.

Tools like containers and Kubernetes have made applications easier to move, but they have not solved the same problems for data or governance. True modernization must go further. It is no longer enough to measure success by how many systems you upgrade or migrate. Modernization now means building smart, unified data ecosystems that keep information usable and governed no matter where it lives. When data is the foundation, everything else becomes easier: governance, adaptability, security and execution all improve.

AI makes this even more important. Early AI tools worked inside single systems. Today’s advanced AI coordinates across many systems and data sources. Without adaptable, well-governed data, AI investments of any size and scale may become fragmented and inefficient.

To close the gap, public sector infrastructure IT leaders should make four key shifts:

Design for exit as carefully as for entry. Every major workload decision should include documented analysis of relocation costs, downtime and risks.
Make data the primary design priority. Shift the central question from workload placement to whether data can be accessed, governed and used across environments.
Don’t box yourself in. Avoid over-concentration in any single environment to maintain viable execution paths.
Measure switching costs explicitly. Track the real expense of relocating representative workloads and data.

Turning AI into sustained advantage

Public sector organizations can finally achieve real modernization by accepting that the old approach belongs in the past. The traditional playbook of moving workloads and measuring progress by migration counts has been overtaken. AI has completely upended modernization. What matters now is building systems where data remains dependable and immediately usable no matter how conditions change.

By putting data at the center of modernization, public sector organizations can move beyond scattered pilots and experiments. They can turn AI capabilities into reliable, sustained outcomes at scale.

]]>

AI won’t replace human claims processors, VA officials say, as Democrats worry over staffing levels

Christian Robles — Tue, 14 Jul 2026 10:47:00 -0400

Artificial intelligence tools the Department of Veterans Affairs is using to speed up disability claims processing will not replace human reviewers, agency officials told lawmakers Monday, but Democrats contend claims processing also need sufficient staffing levels, not just new technology.

“Every disability claim is decided by a trained VA employee, not by AI or automation. These tools support human decision making; they do not replace it,” Robert Orifici, acting deputy chief information officer at VA’s Office of Information and Technology, said during a House Veterans’ Affairs Technology Modernization Subcommittee hearing.

“What we really strive to do is look at those really high volume, highly administrative tasks, focus our resources and [apply] really basic automation on those capabilities, so that we are really supporting and augmenting the productivity of our employees,” added Derek Herbert, acting chief production officer at VA’s Veterans Benefits Administration.

Rep. Tom Barrett, R-Mich., who chairs the subcommittee, expressed support for VA’s focus on maintaining human disability claim reviewers, saying Congress “will make sure that there are strong guardrails where no veteran claim is fully decided by AI or automation.”

The comments come as VA is actively using AI to speed up disability claims processing by prepopulating toxic exposure memos veterans must file, streamlining information gathering and facilitating other procedures. VA is also planning to acquire or develop additional AI and automated tools for disability claims, Nextgov/FCW reported in March.

But Rep. Nikki Budzinski, D-Ill., the panel's ranking member, argued that the new technologies may actually hinder VA reviewers’ ability to quickly process claims.

“I hear from employees that the so-called fixes that VA has pushed, like automation and pilots of artificial intelligence tools, often produce incorrect information, compounding issues and slowing production,” Budzinski said.

The congresswoman said new technology cannot solve declines in VA staffing, adding that she is alarmed that VA has lost 1,100 claims examiners this fiscal year, along with 2,700 examiners and 120 IT specialists since January 2025.

Rep. Maggie Goodlander, D-N.H., said she shared Budzinski’s concerns that a lack of adequate VA staffing is leading to less accurate reviews of claims.

Goodlander pressed Hebert to reveal how many claims examiners the VA has on staff and confirm the total has declined since January 2025. Hebert did not have specific numbers but said he would get those to the lawmakers while noting that VA is currently hiring. VA had several open claims examiners positions on usajobs.gov as of Tuesday morning.

Sterling Thomas, the Government Accountability Office’s chief scientist, similarly told lawmakers that federal agencies need a workforce with the right skills to harness new technologies. He argued that solutions such as a federal digital services academy could help VA and other agencies field the talent they need.

Thomas also recommended that VA turn to GAO’s AI framework of key practices as it continues to roll out new AI and automated technologies for claims processing. The framework says federal agencies should document the data used to build AI models, articulate system-level technical specifications, continuously monitor AI performance, assess that performance and take other measures to ensure technology is responsibly deployed.

]]>

AI, once relegated to helping hackers with certain tasks, can now power every stage of a cyberattack

David DiMolfetta — Mon, 13 Jul 2026 21:00:00 -0400

Just two years ago, hackers were tapping into generative artificial intelligence to probe targets, translate technical material and troubleshoot malicious code. The technology sped up some key parts of a cyber operation, but other stages remained solely in human hands.

That line is now beginning to blur. In a range of cyberattacks observed over the past year, AI systems generated commands, tested vulnerabilities and helped hackers move through victim networks, sometimes carrying out thousands of commands with less human direction than researchers had previously seen, according to research released Monday night by cybersecurity firm Check Point.

It means AI has now been used in some form at every stage of a cyberattack, from identifying targets to exploiting vulnerabilities to stealing data, to help hackers achieve their goals.

The shift does not yet amount to fully autonomous hacking, a top company executive told Nextgov/FCW, but it shows that AI is moving from an occasional aid to an extra set of hands throughout the entirety of an intrusion process. The findings also highlight how rapidly the global cyber ecosystem has adopted AI, with the technology now embedded across every part of an offensive operation rather than confined to isolated tasks.

“We watched criminal groups breach government agencies at scale, using AI as the primary operator rather than a background assistant,” the Check Point report says. In most cases, the AI model’s role was revealed by the attacker’s own mistakes or monitoring by the AI provider, rather than tools or safeguards deployed by the victim organization.

For these AI-enabled pursuits, hackers have used open-source models and purpose-built malicious AI tools sold on the dark web, but major commercial providers remain their primary choice, company threat intelligence lead Sergey Shykevich said in an interview.

The research points to the Gentlemen ransomware group as one instance of how AI is being folded into routine criminal operations. Members compared mainstream commercial models based largely on which imposed the fewest restrictions and used AI to help build internal tools, including a management platform developed in three days.

The findings also highlight VoidLink — a sophisticated toolkit for remotely controlling infected computers — that researchers initially believed had taken a team several months to build. Check Point later found that a single developer produced roughly 88,000 lines of working code in under a week using a commercial AI coding tool.

Hackers tend to first choose U.S. AI models like ChatGPT or Claude because their responses are generally considered higher-quality, but they have a tougher time exploiting those families of models because of stronger guardrails designed to prevent malicious use of the tools, Shykevich said. When those attempts fail, they then pivot to Chinese-made AI platforms that have lower guardrails.

“They are trying to jailbreak those [Western] models, but when they are not successful, they just go to DeepSeek, Qwen and Trae,” he said, referring to a trio of AI models and platforms originating in China. Qwen is a suite of AI and large language models developed by Alibaba, while Trae is an AI-backed code editor and programming platform built by ByteDance. Ransomware gangs have been heavily leaning on those Chinese models to help generate code for their exploits, he noted.

Chinese AI models have recently grown more capable and widely used for coding tasks. Beijing has discussed restricting overseas access to some advanced models, underscoring their growing national security value and their appeal to cybercriminals seeking tools to help with their exploits.

When asked how these new dynamics involve targeting U.S. government and critical infrastructure organizations, Shykevich said there are certainly higher and faster levels of exploitation attempts, though he assessed that’s due to a combination of AI-enabled speed and geopolitical tensions.

He said that, in an ideal world, security flaws should now be patched within several hours of discovery, though he immediately acknowledged that would be next to impossible. The Cybersecurity and Infrastructure Security Agency recently revamped its remediation timeline guidance, ranging from three days for the highest-risk flaws to 60 days for lower-priority issues.

The move is part of CISA’s response “to the current threat landscape where AI software services can assist threat actors to find and exploit vulnerabilities,” the agency said last month.

The findings come as the latest tranche of frontier AI models is showing sharp gains in their ability to perform cybersecurity tasks. OpenAI last week released GPT-5.6, which it called its strongest cybersecurity model yet, reporting substantial gains over its predecessor on benchmarks testing exploit development and proof-of-concept generation.

The Trump administration, meanwhile, has given the National Security Agency, CISA and other federal officials until Aug. 1 to develop a classified process for benchmarking the advanced cyber capabilities of frontier AI models and determining which systems need additional government scrutiny.

“The most significant shift this report documents is not a new technique. It is pace,” the Check Point report says. “A vulnerability now becomes a working exploit within hours of disclosure. Phishing campaigns run at a quality and volume no human team could match. Intrusions span dozens of targets simultaneously, with AI handling the operational work between check-ins. Security teams working at human speed cannot match that cadence.”

]]>

DOD suspends CMMC Phase 2, launches 60-day ‘reform’ review

Nick Wakeman — Mon, 13 Jul 2026 16:55:00 -0400

The Defense Department has essentially ended the Cybersecurity Maturity Model Certification program by suspending its second phase requirements.

DOD is keeping in place Phase 1, which requires self-assessments for how companies protect controlled unclassified information in their systems. But DOD said Monday it is suspending Phase 2, which was to begin on Nov. 10 and requires third-party certifications.

DOD is also launching a review of CMMC to make sure it aligns with Defense Secretary Pete Hegseth’s acquisition initiatives, which prioritizes speed, and lowering barriers for new entrants. The Acquisition Transformation System directives also aim to replace bureaucratic compliance with what DOD calls “scalable, resilient cybersecurity measures.”

The CMMC program got its start during the first Trump administration and was revised and streamlined during the Biden administration. CMMC has been envisioned as a cyber and supply chain security standard for the defense industrial base.

According to DOD's Monday statement, the department is responding to complaints that CMMC was increasing compliance costs and adding bureaucratic burdens.

The Small Business Administration also reported that CMMC compliance had caused some companies to leave the defense industrial base, which DOD said is delaying the deliveries of critical capabilities to operators.

"In support of Secretary Pete Hegseth's directive to reduce compliance barriers for small and medium sized businesses, we are today suspending the CMMC Phase II requirements and initiating a 60-day study of the future of this program," said DOD Chief Information Officer Kirsten Davies.

Davies said that cybersecurity and operational resilience are critical priorities.

“We believe the DIB can achieve both, while we reduce unnecessary government red tape,” she said.

Phase 3 of CMMC, which was to begin in November 2027, and Phase 4 for full implementation are also all suspended.

In the meantime, the department said it would rely on “self-assessments and select government-led assessments.”

DOD's announcement said that Davies made the decision.

“The CIO's decision ensures we maintain a strict security baseline while removing paralyzing costs and keeping innovators and competition growing in the defense supply chain," said Michael Duffey, defense undersecretary for acquisition and sustainment.

DOD has also formed a CMMC Reform Task Force to conduct a review of the certification program. Part of their role will be to review comments in responses to a request for information, which DOD posted Monday.

The department wants feedback from companies on cost drivers, administrative burdens tied to CMMC compliance, and which NIST 800-171 security controls deliver meaningful risk reduction.

DOD also wants to know know how companies are already using commercial cybersecurity tools and managed services, and how the department might recognize those in a compliance framework instead of requiring separate assessments.

Responses to the RFI are due Aug. 14.

]]>

Beeck Center leader reflects on ‘bipartisan opportunity’ for government efficiency

Chris Teale — Mon, 13 Jul 2026 16:23:00 -0400

July 4 meant the end of the Department of Government Efficiency, as the federal initiative launched by President Donald Trump’s January 2025 executive order officially sunset.

DOGE says it has saved billions of dollars, but its record is unclear. What is clear is that the initiative caused chaos at various federal agencies with layoffs and closings, as well as efforts to use technology to solve vexing problems that led to many privacy concerns.

Against that backdrop, and as many agencies start hiring again, state governments also experimented with their own efficiency efforts. Some were high profile, like in Iowa, Florida and Texas, which set up their programs to try and imitate the federal version.

But other states have looked to rethink efficiency and instead define it as the effective delivery of government services. One such state, Arizona, talked up its planned monetary savings but also framed the efficiency effort as a way to rethink certain business processes.

At the forefront of thinking about government efficiency is Georgetown University’s Beeck Center for Social Impact and Innovation, led by Executive Director Lynn Overmann, a veteran of the U.S. Digital Service — which was repurposed to house DOGE via executive order — as well as other federal agencies. She also spent time as a public defender.

In the run-up to GovExec’s Government Efficiency Summit this week in Washington, D.C., Route Fifty caught up with Overmann to discuss where government efficiency goes from here, and how the trend is unfolding at the state level.

This interview has been edited for brevity and clarity.

ROUTE FIFTY: What does efficient government mean to you?

LYNN OVERMANN: I've worked both in the federal government and with state and local governments, so have a pretty broad perspective on it. But at the end of the day, efficient government means that government meets the needs of its constituents with the least amount of burden, both for the constituents and for the government workers that are delivering these services.

ROUTE FIFTY: Days after the end of DOGE, as you reflect back on that, what did you think of it, its aims and its outcomes?

OVERMANN: What was really fascinating to me observing DOGE is that it felt like it was repeating some of the lessons we learned when we first launched the U.S. Digital Service after the Healthcare.gov crash, which was this deep belief that the technology itself was the problem and that therefore technology was the solution. What we learned over time at the federal level — and I think it's being implemented in a far more innovative and effective way at the state level at this stage — is that it's not the technology. You need to understand who your users are and what their needs are, and then you need to build the process, the operations, the regulations, all the things that go around the technology to understand what is standing between the individual who needs the support and the service that's being provided, and then the technology often would come in at the back end.

It was really interesting to watch DOGE come into it with, frankly, a much stronger remit than we had. They were given a lot more air cover, they were given a lot more executive support, they were sent into agencies, and the agencies were told that they had to work with DOGE. Even with that, I think they ran into the same barriers that we did, which is, it’s not a matter of building a snazzy new technology product, it is really deeply understanding how government works and trying to make it better. That is a very different process than building a private sector technology tool.

ROUTE FIFTY: What's happening out there in state government?

OVERMANN: I would name a few states in particular: Maryland, New Jersey, Pennsylvania and Colorado are really doing some really interesting stuff. Colorado, I think, is a particularly interesting example, because their digital service team started not long after the U.S. Digital Service team and actually took some of the alumni from the original USDS and the original White House Office of Science and Technology Policy to launch that team, so they've been around about the same amount of time.

What the digital service team has evolved into under their new leadership is they've recognized that a digital service team on its own, even if it's highly empowered, is not as effective as an overarching technology shop, so have really stepped into the chief information officer shop, which brings with it a lot more responsibility, a lot more authority and a lot more capacity. It’s trying to transform the CIO’s office from a more traditional technology shop into a delivery shop. If I was to name one thing that's really happening at the state level that I think could be an example that the feds should follow at some point, it is recognizing that the goal is delivery, the goal is not necessarily a digital service or a technology tool.

I think we're seeing that in Maryland as well, where they have built out a really strong team and they've embedded folks in the key parts of the state government that not only build technology in partnership with agencies, but also have a lot of carrot and stick authority over the agencies themselves as they're doing procurement. They’re trying to wrap their arms around procurements to ensure not just that they've got a strong internal team that is capable of building a tool, but recognizing that a lot of government technology really comes from procurement, making sure that the procurement is well crafted, that there's a really clear sense of what the tool is that is needed, and then managing the vendors to ensure that it's actually delivered to the public.

ROUTE FIFTY: I want to ask about the goal being delivery. It must be hard as a government, because it must be very tempting to just bolt in a new technology and say, ‘Alright guys, we're done.’ How do you change that? Is that changing a mindset?

OVERMANN: It's both changing a mindset and changing an approach. One of the things that we've talked about for a couple of years now is something called the product model, which does tie back to procurement. It’s everybody's favorite thorny topic that is a challenge for everyone, but is recognizing that buying technology is not the same thing as buying a school bus, for example. The point of the technology is that it is responsive to the emerging needs of the program, and so technology is an ongoing piece of work. It is not a one-time procurement, and I think a lot of the procurements that we've seen — there's a number of reasons that procurements tend to go sideways — but one of them is this belief that you can name what you need the technology to do, you can buy the technology, you can deploy the technology, and that's the end.

I do think the mindset shift that we're starting to see, where delivery and outcomes are the goal, is an ongoing recognition that there needs to be ongoing work. User needs are going to change over time, technology is going to change over time, programs are going to change over time, funding is going to change over time, and what you really need is an infrastructure in place within government. A lot of it comes down to internal talent and capacity, who are capable of understanding in an ongoing way how those needs have changed, and how we should adjust the technology to be responsive in an ongoing way. It's not the way the government typically operates, and it's going to come from change management. That's why it's going to be so fascinating to watch these states that are pioneering in this space move to a goal, a new north star of, this is what they’re aiming for, and how do you bring these government agencies along to actually deliver it?

ROUTE FIFTY: Just on the procurement point, many states talk about having a request for proposals that's out of date by the time it comes into effect. How do we fix that?

OVERMANN: You need to have people on the inside who understand technology well enough to say, “What should these RFPs hold? Do we need an external vendor, or could we actually feasibly build this ourselves?” It's not always an obvious answer. I'm not ever a solid, "You must build," or "You must buy.” I think it depends on what you're delivering and what technology you already have in house, and then if you are going the vendor route, making sure that what you're asking them to do is deliver quickly to demonstrate that they are capable of building and delivering this technology, and then testing it.

It's really funny, we've seen over time the word “agile” has just gotten attached to a bunch of procurements, and, when you look at it, the actual process remains almost the exact same, which is you spend multiple years gathering requirements from different programs, you put them all into one incomprehensible multi-hundred-page document, you get two to three large vendors who respond because they're the only ones who have the capacity to actually dig through all of it and provide a response, and then by the time the project launches you can be two to five years into this process. Your requirements are out of date, and that's not even saying the amount of time it takes for the vendors to then build tools, because you've tried to capture everything all at once in one place. We're still talking four to five years for the tool itself to get developed, or for the system to be deployed, so you're anywhere from five to seven years into a product by the time you get to a point that you're starting to use it. The way technology is moving these days, that is just woefully out of date.

ROUTE FIFTY: Is it also difficult to do any modernization or efficiency project when you've got these old systems that can't break?

OVERMANN: There's been a really interesting project that emerged from the government of Alberta in Canada. They had a bunch of agency systems that were at the end of life, and they were facing a multi-year procurement, I think it was a $50 million RFP, with at best new systems in place in three or five years. What they did — this was a thoughtful use of AI — is the Alberta team had an empowered technology lead who actually understood what agencies’ technology needs were. He had a team underneath him that knew how to actually deploy using AI, so it wasn't just vibe coding to say, “Update my system.” It was, “What is the system architecture we need in place? How are we going to combine systems so that we are reducing redundancy? How are we understanding and meeting the agency's needs, and then deploying his team to use AI to build modularly?”

They were able to modernize multiple of their systems in just eight months, and I think for under a million dollars. That’s an internally empowered team that deeply understands their users' needs, and, in this example, it was an internal government need using modern technology to build things quickly. I think that’s a path that more governments should be pursuing in the future.

ROUTE FIFTY: You’ve talked about change management and having folks on the inside who care about this stuff. How do you go about getting people on board?

OVERMANN: Having clearly defined outcomes and goals. The vast majority of people who work for government are there because they truly believe in the government's mission. Every career staffer I've ever worked with on any technology project, if you can show them that what you are suggesting that they do is different from what they're currently doing and will actually get the job done better and faster, they'll do it. The disconnect is skepticism that the delivery will actually happen because folks have seen lots of great ideas come and go that don't actually work, and I think that's a totally fair concern around whether it’s a flash in the pan or a sustainable approach.

One thing that we learned over time in the federal government certainly is bringing team members in, whether it's programmatic team members or whether it's traditional technology IT shops, and co-designing with them, making them feel as if they're a part of the project from the outset. Nothing gets people on board more quickly than success, and they know the problems way better than we do. Oftentimes, it's not that they don't know what better looks like, it's that they haven't been able to do it with the tools that they've been given. If you bring them in as the experts, and then you help them get where they know they want to be, that will handle a lot of it. If you just parachute technologists in, they come in, they don't actually understand what the delivery challenges are, and they say, “We're going to just deploy an AI tool to solve this in two days that you've been working on for 10 years.” It's not successful, and you're not having a lot of proponents for it when you walk out the door.

ROUTE FIFTY: What's next? What are you going to be paying attention to in this government efficiency space?

OVERMANN: I actually think there's a really interesting bipartisan opportunity coming down the pike. There's a ton of upcoming gubernatorial elections, and many governors who will be changing, either due to term limits or folks just not running again. There's going to be a whole new crop of governors at the state level. There's going to be a whole new crop of state legislators, and they're staring down a tremendous amount of opportunity with emerging AI and other technology opportunities, and they're staring down a huge implementation challenge, if you're just looking at things like H.R. 1.

The fascinating thing to me about state governments that I think is just so different from federal is so many states have balanced budget requirements, and they're forced to make trade-offs that the feds just aren't. Having an empowered, technologically savvy internal workforce and turning them loose on these projects and problems is probably going to generate a whole lot of new examples of how you can actually deliver government in a more efficient way, because you're going to have to in a more constrained environment, but you don't have a choice to not deliver these services.

I think there's going to be a ton of really compelling examples emerging from state governments over the next six to 12 months, and I think that's going to happen in both red and blue states, and I think that's a great opportunity to point out that government should work well, regardless of what your exact philosophy is on what government should be doing.

ROUTE FIFTY: Anything else you’d like to share?

OVERMANN: The civic tech ecosystem has also evolved in parallel with government work, and has gotten stronger and more robust with a lot of the lessons that we learned at the federal level. Us at the Beck Center, Code for America, U.S. Digital Response and others are full of folks who have been government practitioners working on these issues, and provide a really helpful outside space for government partners. We convene our government partners, and we give them a safe and trusted space to share their own problems with each other to shortcut. One of the things we've seen over and over in government is they often have the same challenge, but because they're not necessarily connected, they'll solve the problem themselves instead of understanding that the problem has already been solved.

]]>

Russian hackers exploit weak router security to breach critical infrastructure, Western allies warn

David DiMolfetta — Mon, 13 Jul 2026 15:34:00 -0400

Russian state-backed hackers tied to Moscow’s FSB Federal Security Service are continuing to breach critical infrastructure networks around the world by exploiting routers and other networking devices protected by weak credentials or running outdated technology, the United States and 12 partner nations warned Monday.

The joint advisory said operators from FSB’s Center 16 signals intelligence unit have been working to opportunistically compromise infrastructure organizations across the communications, defense industrial base, energy, financial services, government facilities and health care sectors using the poor router configurations.

The advisory was issued by the NSA, Cybersecurity and Infrastructure Security Agency, FBI and Defense Department Cyber Crime Center, alongside agencies from Australia, Canada, New Zealand and others.

“This is an ongoing issue that has impacted various U.S. and foreign networks across multiple sectors,” the National Security Agency said in a statement accompanying the advisory.

Center 16 operators have also exploited Cisco’s Smart Install feature, web portals used to manage network devices and at least two Cisco vulnerabilities. One of the flaws was logged Monday in CISA’s Known Exploited Vulnerabilities catalog, which tabulates cyber vulnerabilities that have been leveraged by hacker groups.

The guidance comes the same day the United Kingdom and European Union formally blamed Center 16 for a failed December 2025 cyberattack against Poland’s energy grid. UK officials said the operation could have cut electricity to roughly 500,000 people during the winter.

“We strongly condemn Russia’s behaviour and misuse of this cyber ecosystem, targeting public services and critical infrastructure, causing disruptions and financial losses,” EU foreign policy chief Kaja Kallas said.

Former U.S. officials urged organizations to shore up their routers’ cyberdefenses.

“Organizations should treat internet-facing network infrastructure as a priority attack surface by eliminating default credentials, restricting management interfaces, and ensuring that routers receive the same level of monitoring and patching as other critical systems,” said Matt Hartman, the chief strategy officer at Merlin Group who has held several senior roles at the Cybersecurity and Infrastructure Security Agency.

“Organizations should assume that a router or other perimeter device will eventually be compromised. The objective is not simply to prevent the initial intrusion, but to ensure that the compromise cannot spread and create operational consequences,” said Lou Eichenbaum, the former chief information officer at the U.S. Department of the Interior and current federal chief technology officer at ColorTokens.

]]>

DHS network intrusion was twice ruled a false positive before breach confirmed

David DiMolfetta — Mon, 13 Jul 2026 11:24:00 -0400

Department of Homeland Security personnel twice dismissed signs of cyber intruders inside the agency’s Homeland Security Information Network as harmless activity, allowing hackers to remain undetected inside for weeks and eventually steal credential files, according to an internal incident readout viewed by Nextgov/FCW.

HSIN was breached about two months ago, Nextgov/FCW first reported in late June. The network houses sensitive, unclassified data that’s shared between federal, state, local, industry and overseas partner organizations.

Department investigators have still not determined the affiliation of the hackers, according to two people with knowledge of an ongoing probe into the incident. DHS may send staff to brief Congress on the hack in a classified setting in the coming weeks, added the people, who spoke on the condition of anonymity to communicate the department’s thinking.

Between May 15 and May 24, the infiltration was detected by analysts inside FEMA, where they observed the hackers had altered files on testing and live servers, used a legitimate web-server program to run malicious code and deleted activity logs that could have exposed their movements, according to the readout. The activity was ruled a false positive.

Between May 25 and June 3, the hackers used similar methods aiming to leave scant trace of their activity, setting off more alerts that were again dismissed as benign. On June 4, they installed hidden backdoors and stole credential data — typically employed to verify users’ identities and grant access to accounts or systems — where personnel then declared a breach was active.

It’s not clear why the intrusion was deemed benign two times over such a wide timeframe, but the incident highlights how a mistaken assessment can give hackers significantly more time to deepen their access into a target’s environment. The hack involved techniques meant to mask activity as normal, which, generally speaking, can make it very difficult for analysts to determine what is legitimate or not, one of the people said.

It’s also unclear what materials, if any, were copied from HSIN systems, though the fact that hackers targeted credential files indicates they sought out access to accounts or systems beyond what they could initially reach.

"The Department of Homeland Security is aware of a recent cyber incident involving a specific, unclassified legacy information sharing environment," a DHS spokesperson wrote in the same statement it provided earlier this month that confirmed the hack. "We immediately took action to isolate the affected systems, mitigate the vulnerability, and launch a comprehensive forensic investigation. There is no indication that classified networks were impacted, and the system remains operational for our partners. As this is an ongoing investigation, we cannot provide further operational details at this time."

Approved users lean on the network to securely access data, exchange requests with partner agencies, coordinate safety and security for planned events, respond to incidents and share information needed to protect their communities, per its website.

HSIN has been used to support ongoing World Cup games and recent America250 events, Senate Intelligence Committee Vice Chairman Mark Warner, D-Va., said in a statement after the breach was reported.

“The information in HSIN, while not classified, is highly sensitive, and its exposure risks national security,” he said at the time.

As the United States hosts World Cup matches nationwide, the hack could raise questions about whether the intruders gained access to security plans, interagency communications or emergency response plans for one of the world’s most visible sporting events. It’s also possible that World Cup data was not a target.

Nation-state and criminal hackers routinely target U.S. systems to gather intelligence, steal sensitive information and maintain access to government networks. In February, a suspected China-linked breach of an FBI surveillance system likely exposed the phone numbers of people the bureau was monitoring. Last fall, a widespread breach at FEMA let hackers make off with employee data from both the disaster management office and U.S. Customs and Border Protection.

Editor's note: This article has been updated to include a statement from DHS.

]]>

Tech Bills of the Week: Leveraging AI for cancer research; Securing chatbots for underage users; and more

Alexandra Kelley and Edward Graham — Fri, 10 Jul 2026 17:20:00 -0400

This week’s roundup includes several measures that were introduced in the week leading up to the July 4 holiday.

AI to help fight pediatric cancer

Rep. Michael McCaul, R-Texas is looking to leverage AI capabilities to make strides in cancer research in children by introducing the Accelerating Innovation for Kids with Cancer Act on July 9.

The bill seeks to improve federal government coordination to accelerate the use of AI systems and improve diagnoses, treatments, prevention strategies and cures for pediatric cancer.

It specifically tasks the president to work alongside the secretary of the Department of Health and Human Service to appoint a Coordinator of AI Innovation, who would then identify where the federal government can support pediatric cancer research, with a focus on AI.

That coordinator would be responsible for organizing and improving the data infrastructure that supports cancer research to make it ready for AI analyses, as well as improving clinical trial designs to incorporate multimodal data and AI approaches.

Mandatory chatbot safety measures

A bill introduced by Reps. Valerie Foushee, D-N.C., and Greg Casar, D-Texas, adds to the growing body of proposed legislation that asks for developers to add security measures to their generative AI products.

The People-First Chatbot Act, introduced on July 9, seeks to prevent adverse encounters between humans and AI chatbot through instilling strong safeguards. It specifically prohibits tech companies from using data from underage users, —mainly from their chat logs with the product — to further train AI chatbots. It additionally requires companies to disable harmful AI chatbot design features for minors, limits the application of users’ data for targeted ads, gives users the right to access and delete stored chat logs, requires that developers reveal when users are communicating with AI and and mandates that those developers conduct monthly safety assessments, among other provisions.

“The recent tragedies involving AI chatbots emphasize the urgent need for Congress to ensure AI companies and their products are transparent, safe for kids, and most importantly, accountable,” said Foushee in the press release. “The People-First Chatbot Act makes clear that American families deserve control over their personal data, clear disclosure when they are interacting with AI, real protections for children and vulnerable users, and meaningful recourse when AI chatbots cause harm.”

Spotting AI-made content

A bipartisan cohort of House lawmakers introduced a bill July 2 that would set a labeling regime for content created by artificial intelligence software.

Led by Reps. Josh Gottheimer, D-N.J., Tom Kean Jr., R-N.J., and Sam Liccardo, D-Calif., the Spot the Fakes Act requires the Federal Trade Commission to work alongside the National Institute of Standards and Technology to develop a series of technical rules mandating how covered technology companies and AI developers would need to label content generated by an AI tool.

Key to the bill’s mission is to have an AI-generated label be embedded in that content’s metadata.

“Americans deserve to know if what they’re seeing, reading, or hearing was made by a machine,” Gottheimer said in a press release. “This is a commonsense, bipartisan fix — we’re simply requiring that AI content come with a built-in label, so platforms, journalists, and everyday people can tell fact from fabrication.”

New AI application prize challenges

A bill introduced on June 29 seeks to outfit the National Artificial Intelligence Initiative Act of 2020 with prize competition programs to incentivize the development of AI in the U.S. and addresses measurable present-day challenges.

H.R.9506, introduced by Rep. Nick Begich, R-Alaska, amends the landmark National AI Initiative Act of 2020 to include a federal prize challenge. The challenge would focus on developing AI solutions for a myriad of fields, including next-generation algorithm design, chemistry and material science, biotechnology, cybersecurity, energy efficiency, AI safety applications, mechanistic interpretability, quantum computing and more.

The director of the Office of Science and Technology Policy will be tasked with establishing the prize challenge, acting through the National Science and Technology Council and the Interagency Committee. The secretary of Commerce, the secretary of Transportation and the director of the National Science Foundation would be eligible for setting up their own challenge-based acquisitions and other research and development investment initiatives relevant to the bill’s priority technology fields.

Promoting U.S. quantum information sciences and technology

Reps. Mike Lawler, R-N.Y. and Josh Gottheimer, D-N.J., introduced legislation on July 2 to expand the definition of quantum technologies covered by the landmark Export-Import Bank Act of 1945’s Program on China and Transformational Exports, which sets key priority sectors where the government has committed to supporting U.S. manufacturers.

Titled as the Advancing American Quantum Leadership Act of 2026, the bill would change the current “quantum computing” covered by the 1945 law into “quantum information science and technology,” broadening its application to quantum innovations beyond the computing space.

“When it comes to emerging technologies like quantum, the United States must lead, not follow,” Lawler said in a statement. “This legislation would allow for the Export-Import Bank’s ability to support the full quantum ecosystem, helping American companies compete globally, expand exports, and maintain our technological edge over adversaries like China.”

Companion legislation was introduced in the Senate by Sens. Andy Kim, D-N.J.; and Mike Rounds, R-S.D.

AI at the border

A new bill introduced on June 30 by Rep. David Schweikert, R-Ariz., would establish a pilot program for the U.S. Customs and Border Protection to leverage AI solutions at land entry points along the Arizona border. The technology would specifically include an anomaly detection algorithm.

Following introduction, the bill was referred to the House Committee on Homeland Security.

AI for tax fraud detection

Rep. Vern Buchanan, R-Fla., introduced a bill that seeks to help the federal government catch tax fraud with support from AI.

Introduced on June 29, the AI Tax Integrity Act of 2026 would enact an AI Fraud Detection Pilot Program, implemented by the Treasury Secretary, to examine how AI can catch identity theft; fraudulent claims for tax credits, deductions, or refunds by individuals or business entities; and tax returns improperly prepared by third parties.

The program is intended to last between 18 months and two years. The report documenting its performance would be created 180 days after termination of the pilot, created by the comptroller general and submitted to the Committee on Ways and Means of the House of Representatives and to the Committee on Finance of the Senate.

Markey unveils package of AI accountability measures

Sen. Ed Markey, D-Mass. — a vocal proponent of AI guardrails — released an “AI Accountability Agenda” on Friday that includes many of his previously introduced measures designed to mitigate potential harms associated with uses of the emerging capabilities.

The senator’s agenda is built around six priority areas, which include: giving power back to workers; protecting the privacy and safety of children and teens; keeping civil rights safe from AI bias; putting humans first in healthcare; safeguarding against energy and environmental impacts of data centers; and sharing the AI wealth.

Ten bills that Markey previously introduced are outlined in the agenda. These include measures to eliminate algorithmic biases, develop AI privacy safeguards for children — including the Children and Teens' Online Privacy Protection Act — and curtail the use of AI surveillance and automated hiring and firing decisions within U.S. workplaces.

Markey also previously released a discussion draft of a proposal, the Protecting Communities from Data Center Impacts Act, that his office said “would require data centers to receive a certificate from the federal government prior to construction that affirms the data center has met minimum standards for energy, environmental, and economic impacts.” The proposal is included in the agenda.

Although Markey has not yet officially rolled out that measure in the Senate, Rep. Greg Landsman, D-Ohio, introduced the bill in the lower chamber on Friday.

In a statement, Landsman said “our bill makes sure these environmental impacts are studied so communities have the information they need and peace of mind they deserve.”

]]>

Don’t just pick the low-hanging fruit — harvest the whole orchard

Derek Britton — Fri, 10 Jul 2026 16:54:00 -0400

The private sector isn’t just diving into the deep end of the artificial intelligence pool; it’s swimming with the sharks. The promise of dramatic efficiency improvements, accelerated production schedules and perhaps most important, vastly lowered costs, is fueling AI’s meteoric expansion across every industry. But these businesses are also exchanging AI’s potentially exponential rewards for increased tolerance to risk.

Government agencies don’t have that luxury.

Early last month, the General Services Administration published its Elimination, Optimization and Automation (EOA) Handbook, a guide to help agencies eliminate unnecessary tasks, optimize processes worth keeping and automate repetitive ones. The obvious tool to help achieve these goals is AI, and the instinct is to aim AI at easy targets like low-stakes work where mistakes aren’t as costly or as risky.

That’s not entirely wrong. Automating the easy, mundane tasks that are less risky can save tremendous amounts of time and money, but it also keeps more consequential work stuck. And it’s stuck for good reason because AI can’t be explained or defended to regulators, congressional oversight committees, or already AI-weary taxpayers.

The bottleneck in using AI isn’t missing capabilities or the inability to handle volume, but explainability. When AI makes a decision about a benefit, a claim, a case or a security call, someone will eventually need to defend that decision to an Inspector General, a court, or the public.

That’s something that black-box AI just can’t do. In fact, an April 2026 report from the Brookings Institution found that more than 85% of the government’s high-impact AI deployments in 2025 were missing some of the risk information agencies are required to publish.

The most consequential systems already used in the field today can’t account for how they reached their results. For government agencies, results that can’t be explained are results they can’t use.

The answer isn’t to aim low, but to build AI that can answer for itself no matter where you aim. That requires more than a human in the loop; it means workflows must be transparent and explainable using words we can all understand. Every output should point back to specifics: a rule, a record or a verified source that drove it. A human should have the real authority to review, correct, and overrule. And rules should only update when a change is proven, not silently in the background without human intervention.

When AI can account for itself, you can aim at more consequential, higher-stakes work. Export control, for example, is a decision that carries criminal liability if it's wrong. That’s the kind of work you’d never want to hand off to some black box, but a system that shows its work can now take the first pass. This is precisely the kind of rule-bound work the GSA’s EOA handbook says to automate.

There are cautionary tales all around us about why explainability is critical in high-stakes work. Recently, UnitedHealth allegedly used an algorithm to terminate post-acute care for Medicare Advantage patients. But patients who appealed won more than 90% of their cases. The company’s rationale for continuing to use the algorithm? Only 0.2% of patients actually appealed the terminations. But because, as the lawsuit alleges, the humans in the loop couldn’t understand how the decisions were made, and since they didn’t have authority to overrule the algorithm, it was allowed to continue running.

It’s not all bad news, though. There are plenty of examples of how tackling consequential work has real benefits. Take, for example, how the Transportation Security Administration (TSA) implemented new AI-powered scanning technology with the goal of reducing busy work for TSA agents. If agents have more time to focus on complex issues, they can move people through security lines faster and provide help to travelers who need it most. So, when used well, automation doesn’t replace judgment, it frees up more time for it.

Using AI to automate repetitive, low-stakes tasks isn’t wrong, but it does limit the value AI can provide. AI deployments stuck across government agencies shouldn’t be steered around high-stakes, critical, meaningful work. For example, the GSA’s Federal EOA Playbook identifies how AI can leverage Natural Language Processing (NLP) and Document Understanding (DU) to address challenges associated with unstructured data, which has long been a perennial hurdle in providing efficient service delivery.

What is most needed here is for AI outputs to be made defensible and auditable. If agencies get that part right, the consequential work they’ve been avoiding becomes the work they can finally do.

Derek is Senior Vice President of Government at Seekr. He previously served as the Principal Investigator and Research Scholar at the National Security Agency’s Laboratory for Analytic Sciences. Derek earned his B.S. in Operations Research from the U.S. Air Force Academy and a Master’s degree in Operations Research from Southern Methodist University.

]]>

Trump empties out election commission leadership just months before midterms

David DiMolfetta — Fri, 10 Jul 2026 10:31:00 -0400

President Donald Trump removed the remaining members of the U.S. Election Assistance Commission on Thursday, leaving the independent agency without any sitting commissioners just months before the 2026 midterm elections, according to a White House official.

“The President, and head of the Executive Branch, reserves the right to remove individuals that may not be totally aligned with the important task of securing America’s elections and ensuring every legal vote is counted,” the official said in a statement to Nextgov/FCW. “The Slaughter decision gives the President precedence to do so.”

Democratic commissioners Benjamin Hovland and Thomas Hicks were fired, while remaining Republican commissioner Christy McCormick resigned. VoteBeat first reported the dismissals.

The changes leave the commission, which helps oversee federal voting system certification and election administration programs, without the bipartisan leadership responsible for approving some of its most consequential decisions.

The EAC is perhaps best known as a clearinghouse for administering federal election grants and certifying voting systems under the nation’s voluntary voting system guidelines. While those standards are not mandatory, they have become the benchmark many states rely on when purchasing, testing and deploying election technology.

Although the Cybersecurity and Infrastructure Security Agency serves as the federal government’s lead agency for protecting election infrastructure from cyber threats, the EAC has become an important partner in the broader election security and election education ecosystem. The commission works with CISA, state election officials and voting system manufacturers on issues like system testing, certification, security guidance and information-sharing.

“The Administration from the start has been working across all agencies and local partners to safeguard elections from fraud and abuse, and investing in a strong infrastructure to sustain that mission especially in the midterm elections,” the White House official said.

The removals could raise questions about how the agency will carry out some of its statutory responsibilities as federal officials, states and voting system vendors prepare for this year’s midterm elections.

Career staff can continue many of the agency’s day-to-day operations. Still, the absence of commissioners would complicate actions that require a formal vote or approval, potentially slowing decisions involving voting system standards, certifications and other matters.

The White House official did not offer a specific reason for Trump’s action. But the commission has previously declined to change the national voter registration form to require applicants to provide documentary proof of U.S. citizenship, a step Trump called for in a sweeping March 2025 executive order on U.S. elections.

A federal judge later blocked that portion of the order, ruling that it exceeded the president’s authority because the Constitution gives Congress and the states primary responsibility over election administration and oversight. The administration has said it plans to appeal.

The dispute underscored the limits of the White House’s authority over an agency Congress created as an independent, bipartisan commission following disputes during the 2000 presidential election. Under rules set by the 2002 Help America Vote Act, the EAC is led by four commissioners, with no more than two members from the same political party.

It’s unclear when the administration plans to nominate replacements or how long the commission could operate without sitting members. Any new commissioners would require Senate confirmation.

The leadership vacuum comes as election officials continue preparing for the 2026 midterms while confronting cyber threats, foreign influence operations and persistent efforts to undermine public confidence in the security of U.S. elections.

The removals are an “extraordinary step that demands an immediate explanation from the administration and raises profound concerns about political interference in the institutions that support our elections,” said Sen. Mark Warner, D-Va., who serves as vice chairman of the Senate Intelligence Committee.

]]>