Texas Republican proposes AI incident reporting law for dangerous model failures

Rep. Nathaniel Moran, R-Texas, introduced the AI Incident Reporting Act on Thursday with a plan to make artificial intelligence failures visible to federal officials within days, not months. The draft would require developers to report dangerous capabilities, security breaches and safety incidents to the Commerce Department within seven days of discovery, and the most serious cases would then move from Commerce to congressional leadership and relevant committee chairs within 48 hours.

Moran said the measure is a narrow accountability step, not a full rewrite of AI law. “AI is a powerful engine of innovation, and I want to see it flourish, but not without accountability and not without human oversight,” he said. The bill is meant to close the lack of a clear federal mechanism for learning when highly autonomous systems go wrong, while Commerce would decide which models meet the risk threshold for coverage.

The bill would treat several failures as reportable incidents, including models that try to evade human oversight or resist shutdown, theft or unauthorized access to model weights, offensive cyber capabilities and threats involving chemical, biological, radiological, nuclear and explosive harms. It would also cover systems that could autonomously accelerate more powerful AI development.

On June 12, Anthropic said the U.S. government issued an export-control directive requiring the company to suspend access to its Fable 5 and Mythos 5 models by any foreign national. Anthropic said it disabled the models for all customers to comply, said the government did not provide specific details of the national-security concern, and said it believed officials had become aware of a method of bypassing, or “jailbreaking,” the model.

Moran’s push follows earlier congressional interest in the same problem. In September 2024, House lawmakers introduced the AI Incident Reporting and Security Enhancement Act, which directed the National Institute of Standards and Technology to update the national vulnerability database for AI systems and study voluntary reporting of AI security and safety incidents. A January 2025 Center for Security and Emerging Technology report argued that any mandatory regime should capture the type of incident, the severity of harm, technical data, affected entities and the context in which it occurred, with standardized reporting formats to help regulators compare events across companies.

Moran, who represents Texas’ 1st District and serves on the House Ways and Means Committee and the Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party, has already helped drive AI legislation this year through the bipartisan TRAIN Act with Rep. Madeleine Dean.