US Government Forces Anthropic to Pull Fable 5 and Mythos 5 Over Alleged Jailbreak
About This Episode
On June 12, 2026, Anthropic received a US export-control directive citing national security authorities and ordering it to suspend access to its two most capable models, Fable 5 and Mythos 5, for any foreign national. Because it cannot filter foreign users in real time, Anthropic disabled both models for every customer. The government cited a jailbreak method; Anthropic says the demonstration only surfaced a few previously known minor vulnerabilities and disputes that a narrow jailbreak justifies recalling a model used by hundreds of millions.
Our Take
A lab that spent weeks warning no one could act on a frontier-AI emergency just watched the government act on it overnight, by fiat and in the dark, exposing that we have raw national-security power over the most capable models but none of the transparent, verifiable process that would make that power legitimate.
Continue Reading on Unscarcity
Anthropic Pulled the AI Alarm. Who Can Act on It?
Written days earlier about the missing machinery of frontier-AI governance; today flips it, the government acted on Anthropic with none of the pre-negotiated, verifiable, transparent process the piece says doesn't exist.
How Emergency Powers Kill Democracies (And How to Stop It)
Frames Anthropic's procedural complaint, an overnight national-security order with no transparent statutory process maps onto the article's 'Truth Must Be Seen' rule and the danger of emergency authority without limits or auditability.