Tech bills of the week: Ban on AI superintelligence; AI oversight at the Pentagon; and more

Banning superintelligent AI

Sen. Bernie Sanders, I-Vt., introduced legislation Thursday mandating a complete halt in superintelligent AI development, stipulating that the advancement of any system that can surpass human intelligence must permanently cease.

The Ban Artificial Superintelligence Act also temporarily pauses the development and deployment of advanced AI systems until a federal regulator has established safety guardrails.

“The leaders of the major AI companies publicly acknowledge that they do not fully understand the technology and that it is escaping their control,” Sanders said in a press release. “It is irresponsible for society to allow them to move forward and make these products even more advanced. That’s why I am introducing legislation to immediately pause the development of increasingly powerful AI and ban the creation of systems that humanity cannot fully control — at home and around the world. The future of humanity cannot be left in the hands of a handful of Big Tech oligarchs.”

Beyond the ban on superintelligent AI and the pause on advanced AI, Sanders’s bill also creates a new cabinet-level federal agency whose purpose is to safeguard the public from AI risks and hazards, advised by an Artificial Intelligence Advisory Board. The board itself will be composed of AI experts to provide independent technological and scientific advice. 

Penalties for violating the bill, or continuing to develop superintelligent AI or advanced AI systems, would be subject to the “corporate death penalty,” a rare legal punishment where the government strips a corporation of its legal existence. Violators will also be eligible for a maximum of 20 years in prison. 

Oversight for Defense Department AI

On Monday, Rep. Sara Jacobs, D-Calif., introduced a bill that creates a new AI incident and vulnerability reporting program within the Pentagon to help pinpoint and analyze AI risks.

The Defense AI Reliability and Reporting Act calls for a new repository of AI weaknesses across multiple aspects of AI system adoption, including procurement, testing, development and operation, as well as risks stemming from human-machine teaming. 

Reps. George Whitesides, D-Calif., and Nathaniel Moran, R-Texas, cosponsored the legislation. 

“The Pentagon is racing to adopt AI like most other workplaces, but when AI fails at the Pentagon, the consequences can be deadly and could even pull the United States into a war,” Jacobs said in a press release. “That’s why we need a department-wide program to catch and fix AI incidents and vulnerabilities before they turn into catastrophes. Our legislation will help the Pentagon identify risks earlier, learn from AI failures, and help make AI systems safer, more reliable, and more accountable.”

Limiting AI in insurance claim processes

A proposal introduced by Rep. Greg Landsman, D-Ohio, on Tuesday seeks to ensure that AI is not determining which patients receive coverage for medical claims they submit through insurance companies.

Key to the bill — the Doctors Not AI Act — is ensuring that an AI system does not have the final say as to whether an insurance claim is denied or accepted. It allows AI to be used in the claims process, but only in a “supporting role.”

“When a computer system is making decisions about health care, people will get hurt. We have to make sure health care professionals remain the decision makers,” Landsman said in a press release. “If not, folks will be left with massive bills, their health getting worse, or even life-threatening consequences.”

The bill, co-introduced by Reps. Buddy Carter, R-Ga., Kim Schrier, D-Wash., and Tom Barrett, R-Mich., requires that claim denials based on medical necessity be made by a licensed health care professional. It also mandates disclosure of AI usage in the claims process for administrative staff and patients alike and helps protect access to mental health care for insurance policyholders. 

Monitoring self-improving AI systems

A bipartisan duo of House lawmakers introduced legislation on Monday that would require the National Institute of Standards and Technology to monitor how advanced artificial intelligence tools are being used to drive further AI research and development.

The proposal, from Reps. Pat Harrigan, R-N.C., and George Whitesides, D-Calif., would allow NIST to assess and impose requirements on certain memorandums of understanding related to AI systems.

This includes assessments of the “trends regarding the capabilities of artificial intelligence systems, including such a capability to autonomously facilitate or conduct research on or development of an artificial intelligence system, that might contribute to scientific research and development breakthroughs.”

The lawmakers’ measure comes after an incident in which an advanced AI model from OpenAI escaped from its testing environment and hacked into services operated by Hugging Face. In a press release, Harrigan and Whitesides said their bill directly responds to a July statement from more than 1,000 employees of AI firms that called for federal support of an effort to deliberately pace the development of automated AI systems.

"Just in the past few months, we have seen AI systems break out of their testing environments and threaten the security of entire systems,” Whitesides said in a statement. “When you take humans out of the driver’s seat and let machines train machines, small glitches can snowball into major security risks fast. Today, Washington is essentially flying blind on how fast this handoff is happening. This legislation ensures that when frontier labs partner with federal evaluators, we understand who is actually training and improving these models.”

Strengthening the cybersecurity of small businesses

Rep. Tony Wied, R-Wis., introduced legislation on Thursday that looks to help small businesses bolster their cybersecurity by directing the Small Business Administration to work with the Cybersecurity and Infrastructure Security Agency to offer information on cybersecurity best practices.

The measure would also require that the SBA coordinate with the Pentagon to better share information with small businesses “seeking to enter into a contract with the Federal Government, or seeking to be a subcontractor on a Federal contract, regarding the Cybersecurity Maturity Model Certification program (or successor program).”

The bill’s introduction comes after the Pentagon announced in July that it was suspending CMMC phase II requirements while a task force reviews the program.

“In today’s rapidly evolving digital world, it is critical for small businesses to maintain the highest-quality and most up-to-date cybersecurity infrastructure,” Wied said in a statement. “However, small businesses shouldn't be forced to spend hundreds of thousands of dollars to obtain a level of cybersecurity they don’t need simply because Washington has failed to provide clear rules."