Trump reverses stance on AI safety testing after Mythos concerns
Original: Spooked by Mythos, U.S President suddenly realized AI safety testing might be good | U.S President forced to admit Biden was right on AI safety testing.
Why This Matters
Marks significant policy reversal on AI regulation amid growing concerns over advanced model capabilities
Trump administration signed AI safety agreements with Google DeepMind, Microsoft, and xAI after initially dismissing Biden-era policies. The reversal follows Anthropic's decision not to release Claude Mythos model due to cybersecurity risks.
The Trump administration reversed its position on AI safety testing, signing agreements with major tech companies for government evaluations of frontier AI models. Trump had previously dismissed Biden's voluntary safety checks as overregulation and rebranded the US AI Safety Institute to Center for AI Standards and Innovation, removing 'safety' from the name. The policy shift occurred after Anthropic announced it would not release its Claude Mythos model due to fears bad actors could exploit its advanced cybersecurity capabilities. White House National Economic Council Director Kevin Hassett indicated Trump may issue an executive order mandating government testing of advanced AI systems. CAISI has completed about 40 evaluations, including unreleased frontier models, often gaining access to systems with reduced safeguards to assess national security risks.