In an effort to address growing anxieties over the potential threats posed by advanced artificial intelligence, Representative Josh Gottheimer is set to introduce a groundbreaking bill. This legislation will require certain AI companies to have their new models reviewed by the government for risks related to national security, critical infrastructure, cybersecurity, and bioterrorism.

The move comes as Washington grapples with the implications of powerful AI technologies, such as Anthropic’s Claude Mythos, which experts fear could amplify existential threats. These include the potential misuse of AI by malicious entities to develop superviruses or deadly bioweapons, posing unprecedented global risks.

Gottheimer's bill aims to establish a mandatory vetting process, a measure he believes is crucial following recent evaluations of emerging AI models. This initiative parallels ongoing bipartisan efforts in the House to formulate federal regulations for AI, as well as a voluntary vetting framework considered by the White House.

The urgency of the situation was underscored last Friday when the Trump administration implemented export controls on Anthropic's latest AI models, citing significant national security concerns. “The threats identified from models like Anthropic's Mythos have underscored the critical importance of a government-led review process for advanced AI models,” Gottheimer explained to POLITICO.

As a moderate dealmaker, Gottheimer has been actively working with both Democrats and Republicans to establish a national framework for AI. He currently co-chairs a new Democratic commission tasked by House Minority Leader Hakeem Jeffries with forming the party’s official AI policy agenda.

This commission recently criticized a bipartisan discussion draft on AI, introduced by Representatives Jay Obernolte and Lori Trahan, for not sufficiently addressing the critical issues at hand. The draft proposes overriding some state AI laws and mandates that top developers disclose the safety and security risks of their models. It also suggests involving the Center for AI Standards and Innovation to support voluntary evaluations.

Gottheimer's proposal is currently under review by the House Legislative Counsel to ensure its consistency with existing laws. Meanwhile, he continues to engage with lawmakers across the political spectrum to garner support for his forthcoming bill, which represents one of the most assertive legislative attempts yet to mitigate the potential dangers of rapidly evolving AI technology.