<body><p>STORY: The Trump administration is asking top AI companies to submit their most powerful models for cybersecurity testing.</p><p>The testing would be voluntary and happen before the models are released to the public.</p><p>This comes as security fears mount in Washington over powerful new AI systems such as Anthropic's Mythos.</p><p>:: Archive</p><p>President Trump signed an executive order directing several federal agencies to reach agreement with developers over such tests.</p><p>Under the order, U.S. agencies will have up to 30 days to test the models before public release.</p><p>The move signals a shift in Trump’s AI strategy, from a fully hands-off approach to more active government oversight on national security risks.</p><p>Google, Anthropic and OpenAI have welcomed the order.</p><p>:: Archive</p><p>OpenAI CEO Sam Altman said it “gets the balance right” by developing the best models while keeping them safe and strengthening cyber defenses.</p><p>The order also aims to protect critical infrastructure such as banks, hospitals and emergency services by scanning for cybersecurity weaknesses.</p></body>
