Congress is Working on Artificial Intelligence Safety Rules

Moving beyond the process of companies self-regulating in artificial intelligence technologies, the U.S. Congress is preparing bipartisan bills that regulate cybersecurity, chip security, incident reporting, and catastrophic risks.

While debates on artificial intelligence regulations in Washington have largely revolved around the White House and corporate self-regulation, more quiet yet comprehensive legislative preparations are underway in Congress. The bills, which mostly enjoy bipartisan support, cover personal safety, model testing and certification, military applications, and artificial general intelligence goals.

Under the National Defense Authorization Act (NDAA), a steering committee is being established to examine artificial general intelligence efforts, while steps are being taken to exclude Chinese-origin DeepSeek and untrusted artificial intelligence systems. Additionally, the Chip Security Act aims to introduce location verification for artificial intelligence chips.

While the FRONTIER Act introduces risk management, auditing, and incident reporting mechanisms, another bill being worked on by Senators John Thune, Amy Klobuchar, and Ted Cruz aims to impose a legal obligation on developers to mitigate nuclear, biological, and cyber risks. The AI Incident Reporting Act, introduced in June, makes it mandatory to report safety incidents to the Department of Commerce.