100 cybersecurity experts protest against the suspension of Claude Fable 5 and Mythos 5

At the end of last week, Washington issued Anthropic a sweeping injunction. The Trump administration asked the startup to disable Claude Fable 5 and Mythos 5 for all foreign nationals, “inside or outside the United States.”Faced with this unexpected demand, included under the United States’ highly restrictive export control regime, Anthropic chose to cut off access to the two AI models worldwide. All subscribers to Claude plans are therefore deprived of the two new AI models.

It is rumored that the administration’s injunction stems from a warning issued by Amazon, whose researchers reportedly managed to jailbreak generative AI models. Amazon’s researchers used a very simple linguistic manipulation technique called rephrasing. After the model refused to “review the code for security issues”, they simply asked it to “fix this code”. Fable 5 then agreed and revealed minor vulnerabilities that had already been publicly documented. This prompted pressure from the United States, which fears that China may exploit Anthropic’s innovations.

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

100 cybersecurity experts step up to defend Anthropic

The revolt in Washington has sparked an outcry worldwide. While European policymakers are concerned by the Trump administration’s latest power play, around 100 cybersecurity experts have published an open letter demanding that the ban be lifted. The letter argues that disabling the two AI models amounts to sabotaging the work of cybersecurity defenders.

You can read the letter and add your name at https://t.co/dhhZJSQHaf

— Alex Stamos (@alexstamos) June 15, 2026

For Katie Moussouris, CEO of Luta Security and signatory to the open letter, the vulnerability uncovered by Amazon is not an AI jailbreak. She also points out that other frontier AI models, including OpenAI’s GPT-5.5, can reveal identical vulnerabilities without any circumvention. There is therefore no reason to legislate to restrict the use of Anthropic’s new tools. Anthropic agrees, noting that the disclosed vulnerabilities were minor and already publicly documented.

I’ve seen the paper. It’s not a jailbreak. It was Defense Oriented Prompting (DOP), capabilities defenders need. My thoughts on the hasty Export Controls that made Anthropic pull Fable. If Nat defense is the goal, this just scored an own goal against us

https://t.co/DfiAKEylSV

— Katie🌻Moussouris (she/her/she-ra/she-hulk) 🪷 (@k8em0) June 13, 2026

The letter argues that Fable 5 and Mythos 5 do not offer capabilities truly superior to what is already available elsewhere. Competing platforms and open-source alternatives already make it possible to detect vulnerabilities, as do the new Claude models. According to the signatories, deactivating the new Claude models has “cut defenders off from the best models, created uncertainty in the market, and put American AI leadership at risk without any real risk.” By raising the spectre of a national security risk, Washington succeeded only in punishing researchers and defense experts. Moreover, Washington’s decision could ultimately give China a competitive advantage, whose researchers have not been deprived of their cutting-edge models.

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