The export controls come off and Fable 5 returns
On 30 June 2026 the export controls on Claude Fable 5 and Mythos 5 were lifted, and Fable 5 returned to users worldwide the next day. Access to Mythos 5 for a set of US organisations had been restored earlier, following the government's approval of 26 June.
Why it matters
Eighteen days is how long the first ban on access to an already deployed frontier model lasted. The resolution did more than return the models: it exposed that the industry had no shared way to judge how serious a given bypass of a model's safeguards is, and Anthropic began building one with Amazon, Microsoft, Google and other Glasswing partners.
On 26 June 2026 Commerce Secretary Howard Lutnick wrote to Anthropic's chief compute officer Tom Brown that appropriate safeguards were in place to permit certain trusted partners to access Mythos 5. The permission covered more than a hundred named US organisations and companies, including their non-American employees and Anthropic's own, who had been caught by the original ban. Fable 5 was not addressed in that directive. On 30 June the export controls came off both models. Fable 5 became available the following day, 1 July, to users globally on the Claude Platform, Claude.ai, Claude Code and Claude Cowork; for Pro, Max, Team and select Enterprise plans it was included within up to 50 per cent of weekly usage limits through 7 July, and by usage credits after that. The company said it would re-enable access on AWS, Google Cloud and Microsoft Foundry as quickly as possible. Two undertakings came out of the episode. The first is a shared industry framework for judging the severity of a jailbreak, which Anthropic is developing with Amazon, Microsoft, Google and other Glasswing partners; by its account the absence of such a standard is what made it impossible to communicate the level of risk consistently to government and industry. The second is deeper collaboration with the US government on pre-release testing, information sharing and joint research.