US Moves to Mandate Reporting of Critical AI Incidents
A Republican legislator suggested a law that would mandate AI model creators to disclose harmful features, security violations, and safety events.
The proposed bill, put forth by U.S. Representative Nathaniel Moran from Texas, would require AI companies to inform the U.S. Commerce Department within a week upon identifying hazardous activities, with the Department obligated to alert Congress within 48 hours of the most critical occurrences.
"It's a bill aimed at early detection and alerting," Moran mentioned in an interview regarding the AI Incident Reporting Act.
The legislation arises as increasingly potent AI models elevate risks to national security and public safety. On June 12, the Commerce Department acted against Anthropic's new models for national security reasons, leading to Anthropic shutting down global access to them. The government’s order revealed a lack of a clear framework to regulate frontier AI.
Reportable actions under the proposed legislation encompass a model designed to avoid human oversight, bypass protections, and otherwise weaken the capacity of human operators to manage the model. It also encompasses unauthorized access to model weights that influence a machine's decision-making, along with chemical, biological, nuclear, and other hazards to public safety.
Also Read: How Tensions Grew Worse between Elon Musk and Donald Trump
The proposed bill is the most recent AI regulation introduced in Congress, which has faced challenges in enacting laws amidst discussions on whether federal laws ought to override state laws and if safeguards might hinder innovation and U.S. competitiveness with China.
Also Read: US Tariffs on India an Economic Boon?
Earlier this month, two representatives in the House put out a draft for extensive AI legislation titled the Great American Artificial Intelligence Act, which mandated reporting significant safety incidents to the Commerce Department.
Moran stated that his more focused strategy might lead to a faster route to legislation and believed it would garner bipartisan backing swiftly.
Also Read: How US–China Tariffs Are Disrupting Textile Trade
"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," says Mark Beall, president of the AI Policy Network, who supports Moran's proposed legislation.