The Trump Administration Is Lifting Its Export Controls on Anthropic’s Mythos and Fable AI Models

Key Points:

• Anthropic executives met with White House officials but remain divided over the risks posed by their AI model Claude Fable 5, which the company has taken offline following a government order citing concerns about jailbreaking vulnerabilities.
• The Trump administration's approach to AI regulation is fragmented, with previous efforts to regulate AI stalled and internal disagreements hindering cohesive policy development.
• The White House has requested OpenAI to delay the release of its GPT-5.6 models, reflecting broader government caution toward advanced AI systems amid fears of misuse and security risks.
• Despite government crackdowns, experts warn that advanced AI models capable of sophisticated hacking are inevitable, underscoring ongoing challenges in regulating AI technologies.
• Anthropic has adapted its AI offerings by providing a restricted version of Claude Mythos 5 to trusted partners and a safer public version, while also facing challenges enforcing geolocation restrictions as users in China find ways to bypass controls.