Anthropic Lifts Export Restrictions on Claude Models, Tightens AI Safety Measures, and Deepens US Government Ties
Summary
Anthropic lifts export restrictions on its Claude AI models, rolls out sweeping new safety measures including a 24/7 jailbreak monitoring team, and deepens US government ties as CEO Dario Amodei urges Congress to pass AI safety legislation before adversaries like China can exploit weaker guardrails.
Key Points
- The US lifts export restrictions on Anthropic's Claude Fable 5 and Mythos 5 models, with Fable 5 now available globally and Mythos 5 access restored to US organizations as of June 26, 2026.
- Anthropic strengthens its safety measures by blocking a jailbreak method in over 99% of cases, launching a 24/7 jailbreak monitoring team, and partnering with Amazon, Microsoft, and Google to develop a consensus framework for scoring AI jailbreak severity.
- Anthropic deepens its ties with the US government by committing to pre-deployment testing, early model access for officials, and rapid jailbreak reporting, while CEO Dario Amodei urges Congress to pass AI safety legislation before adversaries like China exploit weaker guardrails.