The US Government Just Pulled Fable 5: What Happened

Last updated: June 13, 2026

Official Sources#

The Short Version#

At 5:21pm ET on June 12, Anthropic received an export control directive from the US government ordering it to suspend all access to Fable 5 and Mythos 5 by any foreign national, inside or outside the United States, including Anthropic's own foreign national employees.

There is no practical way to enforce a nationality check on an API key. So Anthropic shut the models off for everyone.

All other Anthropic models keep working. Opus 4.8, Sonnet 4.6, Haiku 4.5: unaffected.

What the Government Claims#

The directive cites national security authorities. Per Anthropic, the letter did not include specific details. Their understanding is that the government believes it found a jailbreak for Fable 5.

Anthropic reviewed a demonstration of the technique and pushed back hard. Their characterization:

• The jailbreak is narrow, not universal.
• It was used to find a small number of previously known, minor vulnerabilities.
• The "jailbreak" essentially consists of asking the model to read a codebase and fix any software flaws.
• Other publicly available models, including OpenAI's GPT-5.5, can find the same vulnerabilities without any bypass. Defenders use this capability every day.

Anthropic's blunt assessment: "we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

If you want the quick tour of what Fable 5 actually is and what it could do before today, here is our 7-minute walkthrough:

The Irony Nobody Is Missing#

Anthropic has spent years arguing that governments should be able to block unsafe AI deployments. Dario Amodei said it publicly. The company restated it in this very announcement, with a caveat: that power should come from "a statutory process that is transparent, fair, clear, and grounded in technical facts."

What they got instead was a letter at 5:21pm on a Friday with no specific details, based on what Anthropic says is verbal evidence of a narrow jailbreak.

The Hacker News thread hit 400+ points within the hour, and the top comments split into two camps: "they fear-mongered their way into this" and "this is the executive branch lashing out at a company that would not bend the knee." Both can be true.

Adding to the noise: within hours, jailbreaker Pliny (@elder_plinius) claimed a coordinated effort had already broken Fable 5's safeguards across multiple restricted categories. We cannot confirm any link between that and the government's directive. We dig into whether it is connected, and four theories for what is really going on, in why the government pulled Fable 5.

If Fable 5 Is in Your Stack#

Practical steps, in order:

1. Check your model IDs. Anything pinned to claude-fable-5 will start failing or silently rerouting. Several HN users report Fable still responding, which suggests a phased shutdown or silent fallback to Opus. Do not assume the model behind your endpoint is the one you asked for.
2. Fall back to Opus 4.8. It is the closest unaffected model and Anthropic confirmed everything below Fable keeps running.
3. Abstract your model layer if you have not already. This is the second time in a year that builders learned a model can disappear overnight. Vendor-pinned prompts and model-specific behaviors are now operational risk, not just tech debt.
4. Watch for restoration. Anthropic says it believes this is a misunderstanding and is "working to restore access as soon as possible." More details are promised within 24 hours.

What We Still Do Not Know#

• Which agency issued the directive and under what specific authority.
• Whether US citizens could legally retain access (the order targets foreign nationals; Anthropic disabled it for all because verification is impossible).
• Whether this becomes routine for every frontier model release, or stays a one-off.

We will keep this updated as Anthropic publishes details. The bigger questions, why this happened and what it means for everyone building on closed models, get their own posts: why the government pulled Fable 5 and what model dependency risk looks like now.

FAQ#

Why did Anthropic disable Fable 5 for everyone?#

The export control directive Anthropic received at 5:21pm ET on June 12 barred access to Fable 5 and Mythos 5 by any foreign national, inside or outside the United States. There is no practical way to enforce a nationality check on an API key, so Anthropic shut the models off for all customers.

Which Claude models still worked during the suspension?#

All other Anthropic models kept working. Opus 4.8, Sonnet 4.6 and Haiku 4.5 were unaffected, and Opus 4.8 is the closest fallback for anything that was pinned to Fable 5.

What jailbreak did the government cite?#

Per Anthropic, the letter included no specific details. Anthropic's understanding is that the government believed it found a jailbreak, which Anthropic describes as narrow, used to find a small number of previously known minor vulnerabilities, and essentially consisting of asking the model to read a codebase and fix its flaws. Anthropic says GPT-5.5 can find the same vulnerabilities without any bypass.

What should I do if Fable 5 is in my production stack?#

Check anything pinned to claude-fable-5, since calls may fail or silently reroute to another model. Fall back to Opus 4.8, abstract your model layer so a single model disappearing overnight is not an outage, and watch for Anthropic's restoration updates.

Continue Reading#

• Anthropic Now Watermarks All Claude Output: Text Watermarks and C2PA for Files