WASHINGTON — Sens. Maria Cantwell of Washington and Mike Rounds of South Dakota are calling on the Trump administration to take immediate steps to address cybersecurity risks from advanced artificial intelligence, arguing that federal agencies should not wait for Congress to act. In a letter sent Tuesday, Oct. 13, to Director of National Intelligence Jay Clayton and White House science and technology director Michael Kratsios, the senators said recent security incidents should prompt a broader government response. They wrote that advanced AI agents had escaped testing environments, bypassed safeguards and accessed real-world systems without authorization. The letter did not identify specific incidents. Cantwell, the top Democrat on the Senate Commerce Committee, and Rounds, a Republican who serves on the Senate Armed Services subcommittee focused on cybersecurity, said the government should strengthen AI safety and security standards at the National Institute of Standards and Technology. Those standards, they argued, should account for the possibility that malicious actors could use advanced AI to conduct sophisticated cyberattacks, automate attacks at scale or threaten critical infrastructure. The senators also urged the administration to expand independent testing through the Center for Advancing Innovation and Standards for Super Intelligence, the Department of Energy and national laboratories. They said universities, research organizations and independent private-sector evaluators should help build that capacity. A third priority is working with AI companies to develop and deploy tools that can detect cyber threats and shore up vulnerable systems. The lawmakers pointed to existing Defense Advanced Research Projects Agency programs as a way to support cooperation with AI laboratories. The letter asks the administration to bring its full capabilities to bear against the risks. Spokespeople for Clayton and Kratsios did not immediately respond to requests for comment. The appeal comes as lawmakers have struggled to agree on what federal AI safety legislation should include. Congress is in recess ahead of the November elections, leaving little expectation that a bill will advance before the end of the year. Cantwell and Rounds are instead pressing for actions the administration could take using existing authorities and programs. The senators’ recommendations would put federal agencies, research institutions and technology companies at the center of efforts to test advanced AI systems and defend against their potential misuse.