Gottheimer readies AI bill to vet powerful AI models for risk
Rep. Josh Gottheimer is preparing to introduce a bill mandating that some artificial intelligence companies submit their powerful new models to the government to screen for national security, critical infrastructure, cybersecurity and bioterror risks.
It comes as fear grips Washington over new AI models, such as Anthropic’s Claude Mythos, that could turbocharge existential risks posed by the emerging technology — such as enabling bad actors to engineer superviruses or create deadly bioweapons.
Gottheimer's forthcoming legislation, details of which the New Jersey Democrat shared exclusively with POLITICO, would run parallel to a bipartisan effort in the House to craft federal rules governing the technology, and comes as the White House considers a voluntary vetting regime for powerful new models.
Relatedly, the Trump administration decided on Friday to impose export controls on Anthropic’s latest models over national security concerns. Gottheimer told POLITICO that threats identified from models such as Anthropic's Mythos “highlighted how critically important it is that we have a mandatory process for the government to review advanced models".
The coming proposal represents one of the most aggressive attempts yet by a key AI policymaker to mitigate potentially catastrophic risks posed by the fast-moving technology.
Gottheimer, a moderate self-styled dealmaker who has been eager to reach an agreement with Republicans on a national AI framework, currently co-chairs a new Democratic commission convened by House Minority Leader Hakeem Jeffries that’s been tasked with developing his party’s official AI policy agenda.
The commission swiftly blasted the discussion draft from Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) unveiled in June, saying it failed “to meet the enormity of the moment.”
That bipartisan framework would override some state AI laws and require top developers to disclose the safety and security risks of their new models. It also would tap the Center for AI Standards and Innovation — an office within the Commerce Department’s National Institute of Standards and Technology — to support voluntary model evaluations.
Gottheimer added that his proposal is currently under review by the House Legislative Counsel, which ensures a policy is consistent with existing laws, and is speaking with both Democrats and Republicans to rally support.
