Federal working group would study risks from AI-controlled machines

The agency would study safety and security risks and identify voluntary safety practices where appropriate. The work would include risks from combining AI with sensors and robots. The group would work openly, seek input, and impose no new regulatory requirements on private organizations. It could not require disclosure of proprietary information, trade secrets, or classified information.

The agency’s director would publish a report on the group’s work and recommendations. Reports would be available to everyone in written and digital formats. The director could publish updates at least once every two years after the first report. The first report would be due within 540 days after the bill becomes law.