Congress weighs AI security and election-content bills
New House proposals would address foreign efforts to copy U.S. AI models and the use of AI-generated content in politics. One bill would direct NIST to work with industry on voluntary standards against distillation attacks and create a cybersecurity roundtable. Another from Rep. Shontel Brown would require transparency for AI-generated content in political ads and limit AI uses affecting voting rights.
A bipartisan House measure from Reps. Pat Harrigan and George Whitesides would task NIST with crafting voluntary standards and recommended practices alongside private AI developers. It also envisions an AI cybersecurity roundtable linking critical infrastructure operators with technology companies, and it flags Russia, China, Iran, North Korea and related entities as particular concerns in model-copying attacks that use mass prompting.
Separately, Rep. Shontel Brown’s proposal would bar AI use intended to wrongfully deny or defraud someone of voting rights and would require political ads with AI-generated content to carry a disclosure. Rep. Yvette Clarke reintroduced another measure requiring disclaimers when political ads contain AI-made photos or video.
If advanced, these proposals could affect AI developers, critical infrastructure operators, political campaigns, and voters. Voluntary standards may shape security practices without imposing mandates, while disclosure rules could help voters identify synthetic political content. Civil liberties and election administrators may weigh transparency benefits against implementation burdens. The measures’ practical reach would depend on enactment, agency follow-through, and industry uptake.