AI Incident Reporting Act (H.R.9477): requiring a system to report serious AI failures and harms

Fields such as aviation, medicine, and cybersecurity have systems to report and share accidents and harmful events, analyze causes, and prevent recurrence. Near-miss and adverse-event reporting are examples that have turned individual failures into learning for the whole sector.

As AI spreads into important settings, a noted gap is the lack of mechanisms to capture and share incidents, malfunctions, discriminatory decisions, or safety problems when they occur.

H.R.9477, the AI Incident Reporting Act, is a House bill that seeks to address this. As its name indicates, it appears aimed at requiring a system to report serious AI incidents, applying to AI the established idea of reporting and sharing what happened as a basis for preventing recurrence and for policy.

The sponsor is Rep. Nathaniel Moran (R-TX); it was introduced on June 25, 2026, and referred to the Committee on Energy and Commerce. What would count as a reportable incident, and who would report what to whom and how, are provision details beyond the information here, so no firm conclusions are drawn.

Incident reporting is the basic infrastructure that makes visible what is happening, a premise for regulation. AI failures tend to be hidden, and unless shared, the same failures recur. A well-designed reporting system can be a practical way to improve safety without relying on heavy-handed regulation. This bill can be understood as one of the U.S. moves to improve AI safety by learning from what happens.

The facts to hold onto are the bill purpose (a system to report AI incidents) and its current procedural stage (referred to Energy and Commerce).