A Republican lawmaker in the US is proposing legislation that would require artificial intelligence developers to report dangerous capabilities, security breaches and critical safety incidents linked to their models.
The draft legislation, introduced on Thursday by US Representative Nathaniel Moran of Texas, would compel AI companies to notify the US Commerce Department within seven days of discovering dangerous activity involving their systems.
Under the proposal, the Commerce Department would be required to notify Congress within 48 hours of the most serious incidents.
“It’s a catch-it-early and sound-the-alarm bill,” Moran said in an interview about the AI Incident Reporting Act.
The proposed legislation comes as increasingly powerful AI models raise concerns over national security and public safety risks.
On June 12, the Commerce Department took action against Anthropic’s latest models in the name of national security, resulting in the company disabling access to them globally. The move highlighted the lack of a transparent framework governing frontier AI systems.
The bill defines reportable incidents as cases involving AI models attempting to evade human oversight, circumvent safeguards or otherwise undermine the ability of human operators to control them.
It would also cover unauthorised access to model weights, which influence how AI systems make decisions, as well as chemical, biological, nuclear and other threats to public safety.
The proposal is the latest attempt by Congress to establish rules for artificial intelligence, amid continued debate over whether federal laws should override state regulations and whether stricter guardrails could slow innovation and weaken the US in its technological competition with China.
Earlier this month, two members of the House of Representatives released a discussion draft of broader legislation known as the Great American Artificial Intelligence Act, which also included provisions requiring companies to report critical safety incidents to the Commerce Department.
Moran said his narrower and more targeted proposal could move through Congress more quickly and attract bipartisan support.
“No legislation on AI has had much of a chance, but I think there’s a growing demand from the public to see some action,” said Mark Beall, president of the AI Policy Network, who supports Moran’s proposed legislation.
Faridah Abdulkadiri
Follow us on:
