What happened
In June 2026, on June 9, Anthropic launched Claude Fable 5, the first Mythos-class model released to the general public. Until then, models in this class (such as Claude Mythos Preview, launched in April through Project Glasswing) were only available to a small group of cyber defenders because of their ability to find and exploit vulnerabilities.
The same day, the company launched Claude Mythos 5: the same model with some safeguards lifted, restricted to Glasswing partners and, later, selected biology researchers.
Three days later, however, access to both models was suspended by order of the US government.
What Fable 5 brought
- According to Anthropic, state-of-the-art performance on nearly all tested benchmarks, with a larger lead the longer and more complex the task.
- Examples cited: Stripe reported a migration across a 50-million-line Ruby codebase done in a day; the model beat Pokémon FireRed using only screenshots.
- Price: $10 per million input tokens and $50 per million output tokens, less than half of Mythos Preview.
The safeguards
To release the model safely, Anthropic built classifiers that detect requests about offensive cybersecurity, biology and chemistry, and distillation attempts (copying the model). In those cases, Claude Opus 4.8 answers automatically, and the user is told. According to the company, more than 95% of sessions never trigger this fallback.
There was also a privacy change: traffic on Mythos-class models now has mandatory 30-day retention, used only for safety and not for training.
The June 12 suspension
On Friday, June 12, at 5:21 p.m. US Eastern time (6:21 a.m. on June 13 in Japan), Anthropic received an export control directive from the US government citing national security. The order barred any foreign national from accessing Fable 5 and Mythos 5, inside or outside the US, including Anthropic's own foreign-national employees.
Unable to filter users by nationality in real time, the company disabled both models for all customers. Other models were not affected.
According to Anthropic, the directive gave no details; the company understood the government had learned of a possible jailbreak of Fable 5. Anthropic said it reviewed the technique, which found a few simple, already-known flaws that other public models can also find. The company complied but publicly disagreed, saying the government should be able to block unsafe deployments, but through a process that is "transparent, fair, clear, and grounded in technical facts."
The return
On July 1, 2026, Anthropic announced that Fable 5 and Mythos 5 were available again.
What this means for businesses
- Vendor risk is real: a model can disappear overnight due to a regulatory decision. Keep an alternative model configured.
- Avoid relying on a single model for critical processes.
- Regulation of advanced AI is also arriving through export controls, which can affect companies outside the US, such as those in Japan and Brazil.
Sources
- Anthropic, "Claude Fable 5 and Claude Mythos 5" (2026-06-09): https://www.anthropic.com/news/claude-fable-5-mythos-5
- Anthropic, statement on the directive: https://www.anthropic.com/news/fable-mythos-access
- CNBC (2026-06-12): https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html
- NBC News: https://www.nbcnews.com/tech/tech-news/anthropic-suspends-new-ai-models-fable-mythos-government-directive-rcna349901
Want to reduce the risk of depending on a single AI vendor? KGA IT offers a free consultation: kga-it.com.