The U.S. Commerce Department has lifted export restrictions on Anthropic’s top-tier AI systems, clearing the company to restore foreign access to its Claude Fable 5 and Mythos 5 models. Anthropic said access would resume after it agreed to detect and mitigate model security risks, coordinate with Washington on standards for upcoming releases, and report malicious activity. The move reverses a Trump administration directive last month that forced Anthropic to block foreign nationals, including some employees, from using the models amid unspecified security concerns tied to Fable 5. The company had already received limited approval to serve U.S. entities operating critical infrastructure.
The decision eases tensions between Anthropic and the administration following a March lawsuit challenging a Pentagon “supply chain risk” designation. It also underscores a broader shift as Washington tightens oversight of frontier AI, with OpenAI recently opting for a staggered GPT-5.6 rollout under government pressure. Analysts said reported “jailbreaks” of Fable 5 were overstated and warned that blocking one model on such grounds would have set a sweeping precedent. The episode raises a key question for the sector: whether future frontier model launches will require de facto government signoff.
Related articles:
NIST AI Risk Management Framework
The European approach to Artificial Intelligence (EU AI Act overview)
UK AI Safety Summit 2023: outcomes and commitments
OECD AI Policy Observatory
UK AI Safety Institute: mission and research
