The Anthropic Ban Just Eased — For Americans. Europe Is Still Locked Out.

For three weeks, gafam.ai has followed a single thread: the US government locking the world out of Anthropic's most capable AI, the dependency it exposed, and the question of how — and for whom — it would end. On Friday evening, we got the first real answer. And it is exactly as revealing as the ban itself.

What Happened Friday

The U.S. government granted Anthropic permission to release its Mythos 5 model to a group of roughly 100 companies and federal agencies, the Commerce Department detailed in a letter to Anthropic, marking a major step in the two-week standoff over its latest models, Fable 5 and Mythos 5.
The authorisation came directly from the top. "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model," U.S. Commerce Secretary Howard Lutnick wrote in the letter, citing significant progress in the daily talks between the government and the company since the block took effect on June 12.

The mechanism is precise and worth understanding. Under the new arrangement, a licence will no longer be required to export the Claude Mythos 5 model to entities identified in "Annex A" of Lutnick's letter, and to their foreign-national employees, or to Anthropic's own foreign-national employees. In other words: a named list of approved organisations, attached to a government letter, now defines who on earth may use Anthropic's strongest model.

Who Is On the List — And Who Is Not

More than 100 companies and institutions will now have access to Mythos 5, including many Fortune 500 companies, with many of the approved organisations being part of Anthropic's Project Glasswing — the roughly 100 well-known tech companies and institutions that had early Mythos access, including infrastructure providers like Cisco and banks like JPMorgan Chase.
The framing is explicitly American and explicitly security-focused. Anthropic stated that the government notified it that Mythos 5, its strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure.

Note what is absent from every account of the Annex A list: European institutions.

The restoration covers US organisations that defend US critical infrastructure. European companies, European researchers, European agencies and European consumers are not the beneficiaries of Friday's de-escalation. The lockout we documented three weeks ago has been lifted for Americans on a government-approved list — and for no one else.

Fable 5 — The Public Model Stays Dark

There is a second crucial limit. Lutnick's letter is silent on Fable 5, the weaker version of Mythos that was briefly the most powerful AI model widely available to consumers. People close to the talks said they are moving toward releasing Fable as well, though that timeline is unclear.

This distinction matters enormously. Mythos 5 was always a restricted, partner-only model — its return to roughly 100 vetted institutions is a return to roughly its prior state. Fable 5 was the public release — the first time Anthropic made this class of capability available to ordinary users, including in Europe. Fable 5 remains offline. The model that the general public could actually use is still dark, with no timeline for return.

So the practical reality for a European user today is unchanged from two weeks ago: you still cannot access Anthropic's most capable models. The de-escalation was real, but it was for a specific American list, not for the public, and not for Europe.

The Regime This Establishes

The most consequential element of Friday's news is not the partial restoration. It is the system it formalises. Lutnick's letter marks the beginnings of a new regulatory regime that gives the US government control over the release of frontier AI models.

The structure is now visible. Earlier in June, President Trump signed an executive order establishing a framework for AI developers to offer "covered frontier models" to the US government for up to 30 days before releasing them to trusted partners. Friday's Mythos arrangement is that framework in action — and the same day, Anthropic's competitor OpenAI released its newest models, GPT-5.6, only to a small group of government-approved partners under the same regime.
This is the deeper story. Both leading American AI labs now release their most powerful models through a government-approval process. A named list, attached to a Commerce Department letter, determines who may access frontier AI. This is no longer an emergency measure aimed at one company. It is the emerging operating system of American AI distribution.

The European Position — Stated Plainly, By Washington

For once, we do not need to infer the European consequence. It was acknowledged directly in the reporting. The framework for overseeing AI is being built on the fly, and many users of these powerful tools — from non-US governments and companies to consumers — remain in the dark as to when they will get access to Mythos and Fable. European officials and other US allies have expressed frustration at their new dependence on decisions in Washington.
That sentence is the entire gafam.ai thesis, stated by Semafor rather than by us: European access to frontier AI now depends on decisions made in Washington, through a process in which Europe has no vote, no representation and no timeline.

The European Perspective