In mid-June, the U.S. government did something I honestly didn't have on the note:
Anthropic had to abruptly disable its two most powerful AI models after the Trump administration enacts export control directive, which prohibits the use of Fable 5 and Mythos 5 by non-U.S. citizens, even including Anthropics' own foreign employees. The Time Reported on 13.6 similarly astonished about it.
A short list for those who missed it: On 9 June 2026 Anthropic Claude Fable 5 and Claude Mythos 5. Three days later, the U.S. government ordered to block access for foreign citizens. Since Anthropic could not verify the nationality status of the users in real time, both models were simply switched off for all customers worldwide.
The official reason: A jailbreak
The Trump administration instructed Anthropic to immediately block Fable 5 and Mythos 5 from foreign citizens. This apparently happened after Amazon researchers were able to bypass some of the security mechanisms of Fable 5. Specifically, it is a method, which essentially consisted of instructing the model to read a specific codebase and find any security vulnerabilities.
Anthropic itself disagrees with the assessment that this is a sufficient reason for a global ban. The company noted that other publicly available models can also find these vulnerabilities, even without the need to circumvent the safeguards.
76 cybersecurity experts, including CEOs, CISOs and security researchers, In an open letter to Commerce Secretary Howard Lutnick, the decision has taken the best tools out of the hands of defenders, created market uncertainty and weakened America's AI leadership position. To top it all off, even without any noticeable safety gain.
What keeps me busy
I don't want to tick this off as a nasty tech message because it's more than that. In the meantime, Fable 5 was partially available again/still via the mobile and desktop apps. Myth 5 is still closed.. Addendum: Myth 5 was re-released as at 27.6, at least for ‘selected companies’.
But the real problem is another one: A U.S. government agency decided on a Friday night that a tool that people and businesses around the world have paid for and depend on no longer exists. No warning, no transitional period, no voice. Just go away.
This is no longer a theoretical risk. That's what happened. This is exactly what brings me to the following transition:
Digital Independence Day (and why it's not a coincidence)
Ten days before the Fable ban, on 3 June 2026, the EU Commission had its so-called European Technological Sovereignty Package presented. The package includes measures to strengthen EU capacities in semiconductors, AI, cloud and open source with the aim of: Strengthening Europe's digital autonomy and shaping a more sustainable digital future.
Commissioner Henna Virkkunen stressed that Europe should become independent in the digital field and must achieve technology leadership in the long term – and explicitly mentioned US sanctions as well as US export bans on powerful AI models from Anthropic as examples of the risks of existing dependencies.
The timing could hardly have been better to illustrate the point.
The real problem: We're all hanging in there
Anyone who now thinks ‘I don’t use Anthropic, I don’t care’ should think again. The same principle applies to AWS, Azure, Google Cloud, Microsoft 365, GitHub, Cloudflare and anything else based on US infrastructure or US companies. All of this is subject to U.S. law and thus potentially U.S. export controls.
While the EU accounted for 22% of the global ICT market in 2013, it was only 11% in 2022. So we have not only built up a dependency, we have actively deepened it in recent years.
The question is not whether a similar situation can happen again. Instead, the question is only ‘when’; and it remains questionable whether we are better positioned ‘then’.
TL:DR
The fable ban is not an outlier. He's a symptom. The European Technological Sovereignty Package of the EU Commission, consisting of the Chips Act 2.0, the Cloud and AI Development Act, an EU open source strategy and a roadmap for AI in the energy sector, is at least a step in the right direction.
Whether this really becomes something depends on how seriously the member states take it and whether the implementation is not years behind the announcements. As with the EU, unfortunately sometimes It's often like that.
Until then: Anyone who has built critical processes on external, US services should at least have a Plan B in their drawer. Not because the U.S. is evil, but because a Friday night memo is apparently enough to take your entire workload off the net instantaneously.