Tech bills of the week: Preventing AI model theft; Regulating AI-generated political content; and more

Combating AI distillation attacks

A proposal from Reps. Pat Harrigan, R-N.C., and George Whitesides, D-Calif., looks to prevent distillation attacks against U.S.-made AI models by directing the National Institute for Standards and Technology to develop voluntary industry standards and best practices in partnership with private-sector model developers. 

The bill would also create an AI Cybersecurity Roundtable to help foster partnerships between critical infrastructure operators and tech firms to develop additional approaches for enhancing the cyberdefenses of the AI capabilities. The proposal names Russia, China, Iran and North Korea — as well as “entities” tied to them — as actors of particular concern behind distillation attacks, which seek to copy AI models by overwhelming them with millions of prompts to create a knockoff version. 

“Some of the greatest advances in artificial intelligence are happening right here in America, and we have a responsibility to make sure that technology is secure,” Harrigan said in a statement. “Our adversaries are already looking for ways to exploit these systems, and as AI becomes more powerful, the risks to our critical infrastructure will only grow. We need to get the companies developing this technology working directly with the people responsible for defending our power grids, hospitals, and financial systems.”

Guardrails around AI used to affect elections and voting

Rep. Shontel Brown, D-Ohio, introduced a bill on Monday that seeks to promote transparency around the use of artificial intelligence-generated content in political ads and curtail uses of AI to affect Americans’ voting rights.

Brown’s bill would prohibit the use of AI “to intentionally deprive or defraud an individual of the right to vote in an election.” Her measure would additionally mandate that political campaign ads that use AI content include a disclaimer stating: “This output was generated by artificial intelligence.” Rep. Yvette Clarke, D-N.Y., reintroduced a separate bill last week that would also require political ads to include a disclaimer if they include any AI-generated photos or video.

“This isn’t complicated: voters deserve to know that the campaign content they see and hear is real,” Brown said in a statement. “We need labels on AI-generated election content and we need to hold accountable those who use AI to knowingly spread false information. Pairing the latest technology with old-school dirty tricks is a dangerous combination.”

Creating a digital framework of cancer screening guidelines

Reps. Ashley Hinson, R-Iowa, and Hillary Scholten, D-Mich., rolled out legislation on Wednesday that seeks to enhance early cancer detection by making screening recommendations easier to understand for Americans.

The lawmakers’ proposal would accomplish this, in part, by directing that the Department of Health and Human Services “develop a national digital framework for cancer screening and risk assessment recommendations that can be incorporated into electronic health records.” The bill would also authorize new research and pilot projects to enhance the use of new technologies, including supporting the adoption of capabilities that can help healthcare providers “identify patients who may be at higher risk for cancer or overdue for recommended screenings.”

“Early cancer detection saves lives, but for far too many Americans, navigating complicated health care recommendations can lead to missed screenings,” Scholten said in a statement. “I’m proud to join Rep. Hinson in introducing the SCREENS Act, which would help create a national, digital framework of cancer screening guidelines.”