Skip to main content

Washington Bans Fable. Anthropic Wins Anyway

· 4 min read

For the first time, a Western government has forced an AI model off the market. Anthropic disabled its two most powerful systems, Fable 5 and Mythos 5, yesterday after a federal export control directive landed citing national security.

This is a watershed moment. A U.S. administration supposed to fiercely protect domestic tech from overregulation has become the one to pull the ultimate kill-switch.

Anthropic launched Fable 5 on June 9: a version of the much promoted Mythos 5 model with cybersecurity and bioweapons safeguards. Within a day, a hacker who goes by Pliny the Liberator, well known enough to land on TIME's list of the hundred most influential people in AI, bypassed the model's safety guardrails and posted the proof on X. Within hours the post had gone viral and currently has over 1.7m views. He stewards a hacker collective called BT6, a loose network of operators who treat frontier models as locks to be picked. A frontier model had fallen a day after release, as they usually do. Then Amazon, claimed it could jailbreak Mythos 5, Fable's restricted twin. Three days after release, the government ban landed.

The US government directive restricted access to the two models by any foreign national, inside or outside the United States. The company said the only way to comply was to switch the models off for everyone. Anthropic has said publicly that it disagrees with recalling a model used by hundreds of millions over a single exploit, and that the same capability is already available from rival models such as OpenAI's GPT-5.5.

Most of Silicon Valley has read the suspension as score-settling. In February, Anthropic refused to let its models be used for autonomous weapons targeting and domestic surveillance. After an ugly public spat with Pete Hegseth, the Pentagon designated Anthropic a supply-chain risk. A federal judge found Anthropic likely to prove the designation was retaliation, not a security call. The case is on appeal. From that angle, the Fable 5 directive is the same fight in a new arena, with national security authorities standing in for a procurement dispute.

The question worth asking is: who walks away from this stronger?

Anthropic has shown it knows the brand value of restraint. A week ago, The Intercept argued that the earlier Pentagon fight had already worked as a marketing advantage, burnishing Anthropic's image as the principled lab that puts values over profit while rivals chase contracts. A week before the government banned Fable 5, Anthropic published a post arguing the whole industry should be ready to slow down or pause frontier development, citing the risk of models that improve themselves. Critics, including Sam Altman, have long read these warnings as fear-based marketing, a front-runner calling for brakes in a race it is already winning.

Anthropic is drowning in its own success. Anthropic spent much of this year rationing usage, tightening limits as demand outran capacity, then conceded the cause was a compute shortage. The fix is slow. Tens of billions in pledged infrastructure does not become live capacity for 18-24 months. Revenue has roughly quintupled in a year against hardware that cannot keep pace. In February an Anthropic executive told me the company was "just doing everything we can to keep the lights on."

Anthropic can now act the martyr and say its own government forbade it from selling its most capable model, a model "too powerful" for the world from a principled struggling to turn it off themselves. Or it can admit the simpler thing: it is grateful to the government helping to reduce the overall demand. A company heading into an IPO knows which story it would rather tell.

Now layer the IPO on top. Anthropic filed for IPO on the first of June, after a $65 billion dollar round valued it at almost $1tn dollars. Into that window, the government has handed the company a way to blame Anthropic's supply issues on state coercion rather than on the infrastructure problem it has managed all year.

Whether Anthropic is pleased is beside the point. The events line up the same way regardless. The government intended to discipline a company it had been fighting for months. It may instead have written that company the most flattering chapter of its IPO prospectus.