US lawmaker proposes bill to require AI companies to report critical incidents (opens original article in a new tab)

A US lawmaker proposes a bill requiring AI companies to report critical incidents like security breaches and dangerous capabilities to the Commerce Department, aiming to enhance safety and address national security risks.

• US lawmaker Nathaniel Moran proposes AI Incident Reporting Act requiring AI companies to report dangerous capabilities and security breaches to the Commerce Department within seven days.
• Reportable activities include models evading human oversight, unauthorized access to model weights, and threats to public safety such as chemical, biological, or nuclear risks.
• The bill aims to establish a proactive safety framework for frontier AI technologies and comes amid concerns over national security risks posed by advanced AI models.