Anthropic vs Trump, China, Amazon warning… The story behind the suspension of Claude Fable 5 and Mythos 5

Last Friday, Anthropic was forced to disable Claude Fable 5 and Mythos 5 worldwide. The startup found no other way to comply with a U.S. government directive.The Trump administration did indeed demand that Anthropic cut off access to these two models for “any foreign national, inside or outside the United States.” To justify its request, Washington cited national security concerns.

Read also: Anthropic sabotaged the latest version of Claude, and researchers are furious

Why did the United States want to cut off access to the new Claude models?

At the root of the Trump administration’s decision was an alert issued by Amazon. Shortly after the release of Anthropic’s two flagship models, Amazon researchers reportedly managed to bypass several of Claude’s censorship mechanisms. These mechanisms were intended to prevent any malicious use of AI. Amazon researchers were able to demonstrate internally that the new versions of Claude could easily be used to orchestrate cyberattacks.

According to the Wall Street Journal, these findings made their way up to Andy Jassy, Amazon’s current CEO. The executive promptly raised the matter with Trump administration officials, including Treasury Secretary Scott Bessent. It was then that the White House took up the case and asked Anthropic to take action. Amazon, one of Anthropic’s shareholders, remains evasive and merely acknowledges that it is “not uncommon for governments to seek our advice on potential security risks”.

Meanwhile, a report by the media outlet Semafor reveals that a group of Chinese hackers may have found a way to bypass Anthropic’s protections to use Claude Mythos, an ultra-powerful version reserved for a select handful of vetted partners. Considered too dangerous, this version of Claude is indeed restricted to companies that are members of the “Project Glasswing,” a coalition bringing together the biggest names in global tech, including Amazon, Apple, Broadcom, Cisco, CrowdStrike, Google, Linux Foundation, Microsoft, and NVIDIA. According to the outlet, the Chinese hackers, operating with Beijing’s mandate, gained access to Mythos. This is one of the factors that may have prompted Washington to act. The Trump administration fears that hackers funded by foreign powers could use Anthropic’s innovations to identify and exploit vulnerabilities. For context, an investigation showed that a Discord group of curious individuals has already gained access to Mythos by breaking Anthropic’s security.

I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true:

— As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable.…

— David Sacks (@DavidSacks) June 13, 2026

Meanwhile, David Sacks, Donald Trump’s AI adviser, says that a “highly credible” and trusted partner began testing Fable 5 and realized that there was a way to bypass the safeguards.The administration would then have asked Dario Amodei, CEO of Anthropic, to fix the flaw or cut off access to the model. He would have refused to comply, leading to the injunction issued by the government. David Sacks insists that the White House acted “reluctantly”, solely because Anthropic dug in its heels.

He maintains that the measure has nothing to do with the previous disagreements between Anthropic and the U.S. government. A few months ago, Anthropic openly opposed the Pentagon. The firm decided to bar the U.S. military from using its generative AI for military purposes. Anthropic’s rebellion provoked Donald Trump’s anger, who, with his usual emphasis, said that Anthropic was a “FAR-WOKE LEFT-WING COMPANY” that is “run by people who have no idea what the real world is.”

For Anthropic, a “minor potential vulnerability” should not be enough to ban an AI model deployed to hundreds of millions of people worldwide. The king of AI believes that this kind of measure risks blocking almost all new advanced models. The startup presents Washington’s decision as a threat to innovation and progress in AI research. Anthropic’s argument falls a bit flat. A few days ago, the startup did indeed advocate slowing down AI development.

Also read: Biological weapons and AI – OpenAI and Anthropic raise the alarm

A decision that sparks controversy in Europe

The US government’s decision had major repercussions worldwide, including in Europe and France. Several French and European officials reacted to Anthropic’s announcement over the weekend. In the eyes of many elected representatives, the ban announced by Washington highlights just how strong a position the United States holds in the AI industry.

Washington’s decision to cut off access to Anthropic’s most powerful models must wake us up. In the race for artificial intelligence, a nation that depends on others for its technology is a nation that can be cut off overnight.

We have to take action…

— Bruno Retailleau (@BrunoRetailleau) June 13, 2026

For Bruno Retailleau, the former interior minister, the matter must « serve as a warning » and push France to treat AI « as we treated nuclear power », that is, « as a component of our sovereignty ». He advocates redirecting « public procurement toward sovereign solutions » and building infrastructure independent of the United States.The candidate in the 2027 presidential election subsequently even invited Anthropic to establish itself in Paris.

In a context of American disengagement, the European nations must take their destiny into their own hands, work to develop industrial cooperation, and regain the resources needed for their freedom, defense, and strategic autonomy. @POLITICOEurope pic.twitter.com/frU8syX7K5

— Jordan Bardella (@J_Bardella) June 15, 2026

The issue has united all of France’s political parties, from France Insoumise to the RN, including Horizons. Echoing Bruno Retailleau, they believe that it is urgent for Europe to take ownership of AI technologies and not remain dependent on the good will of the United States. Jordan Bardella, the RN leader, believes that «in a context of American disengagement, the European nations must take their destiny into their own hands».

AI: since 2016, LFI has classified the digital realm as a new frontier of humanity and rejects the idea of France becoming a digital colony of the USA. Trump’s decision concerning Anthropic proves the urgency of the issue. It is also, admittedly, a political settling of scores with his…

— Jean-Luc Mélenchon (@JLMelenchon) June 13, 2026

Jean-Luc Mélenchon believes that Trump’s decision looks above all like «a political settling of scores» with Anthropic, but that it is urgent to act to prevent France from becoming «a digital colony of the USA».An observation shared by Edouard Philippe, who believes that “the US government is deciding to subject AI development to its logic of power”.

By suspending non-Americans’ access to Anthropic’s most powerful models, the US government is deciding to subject AI development to its logic of power.

AI is now critical infrastructure, just as essential as electricity or the Internet. A…

— Edouard Philippe (@EPhilippe_LH) June 13, 2026

Among the European leaders, Thomas Regnier, spokesperson for the European Commission, acknowledges that cutting-edge AI raises “serious cybersecurity concerns that must be taken into account”, but that “the emergency measures taken in this context must not discriminate against partners”. He also encourages Europe to “strengthen its technological sovereignty”.

Note that Anthropic hopes to be able to change the American government’s mind. In a blog post, Dario Armodei’s company speaks of a “misunderstanding” and commits to doing everything possible to “restore access as soon as possible”.

👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.