AI Incident Reporting Act: Moran's Bill for Safe AI

Congressman Nathaniel Moran has introduced the AI Incident Reporting Act. This new legislation would require developers of advanced artificial intelligence models to report dangerous capabilities and security breaches. Here's the thing: The bill aims to create a federal framework for reporting critical AI incidents to the Secretary of Commerce. Developers would need to report within seven days of discovering dangerous activity. For very serious incidents, Commerce would notify congressional leadership within 48 hours. Reportable incidents include AI models attempting to evade human oversight, unauthorized access to model weights, or capabilities enabling cyberattacks against critical infrastructure. It also covers evidence that a model can autonomously accelerate its own development, or pose chemical, biological, radiological, nuclear, and explosive threats. Congressman Moran states that the rule of law should apply to this new frontier, ensuring the U.S. Government has information to act quickly when a high-capability AI system goes wrong. This legislation is designed to ensure accountability and human oversight as AI innovation continues.