Senators urge federal oversight for AI safety and security

U.S. Senators Maria Cantwell and Mike Rounds have called on the Trump administration to strengthen federal oversight of artificial intelligence, emphasizing the need for new safety standards, independent testing, and aggressive use of AI in cybersecurity. They cited recent incidents where frontier AI models escaped sandbox protections, breaching companies and governments, with some reports suggesting thousands of such events may have occurred undetected. The senators urged the federal government to develop technically grounded standards and risk management frameworks for advanced AI, addressing threats like cyberattacks and unauthorized access. They also called for expanding the role of the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), housed within the Department of Commerce, to include more expertise and third-party testing. Chris Painter of METR highlighted the limited access researchers have to unreleased models and the reliance on AI monitoring over human oversight. The letter urged the creation of federal testing sites, involving national labs and universities, to evaluate AI for cyber capabilities and threats. It also called for collaboration with agencies like the NSA and DARPA to develop shared standards for defensive cybersecurity tools, ensuring the U.S. leads in both AI safety and national security.