Representative Nathaniel Moran (R-TX) is pushing for the first-ever federal framework for reporting dangerous AI capabilities, security breaches and safety incidents to protect Americans from emerging threats.
The AI Incident Reporting Act would require the United States Department of Commerce to identify high-risk AI models and call on developers to report dangerous incidents within 7 days. For the most serious incidents, such as evidence that a model poses a major public safety risk, the department would have 48 hours to notify members of Congress.
“This legislation ensures that when something goes wrong with a high-capability AI system, the U.S. government has the information needed to act quickly,” Rep. Moran said.
With AI systems becoming more autonomous, Rep. Moran said the government must have a clear and formal route of learning when something goes wrong. This reporting system outlined in the bill does just that, he said.
“AI is a powerful engine of innovation, and I want to see it flourish, but not without accountability and not without human oversight,” Rep. Moran said. “The rule of law should apply to this new frontier.”
The bill, one of Congress’s first major efforts to create a federal framework for reporting dangerous AI incidents, comes a few weeks after President Donald Trump signed an executive order calling for the U.S. to maintain its leadership in AI while prioritizing national security.
Rep. Moran’s bill emphasizes similar priorities, directing the Department of Commerce to work with experts in AI developement, academic research, cybersecurity and national security to ensure the reporting frameworks not hinder these goals.
“The capabilities of advanced AI models are increasing by the day, and so are the risks of AI systems gaining the ability to self-improve, evade human oversight and enabling attacks against our nation’s infrastructure,” said Brendan Steinhauser, CEO of the Alliance for Secure AI, a nonprofit that advocates for AI safety and security.
Steinhauser said early reporting is key to preventing dangerous AI incidents from becoming larger national security threats.
“We need mechanisms to identify these risks early so that leaders can respond quickly before incidents become bigger threats,” Steinhauser said.
