The U.S. government has ordered Anthropic to immediately shut off access to two of its most powerful AI models — Claude Fable 5 and Claude Mythos 5 — citing national security concerns. The decision comes after the company's own safety warnings may have backfired, leading regulators to pull the plug.
According to TechCrunch, the government's order was issued on Friday, demanding immediate action from the AI company.
Anthropic's Strong Disagreement
Anthropic has not hidden its frustration with the government's decision. The company believes the move is an overreaction to a narrow security finding.
"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." — Anthropic via TechCrunch
The company's blog post makes clear that it views the government's action as disproportionate. The models in question are used by hundreds of millions of people, and Anthropic argues that a single potential jailbreak — a way to bypass safety features — does not justify a full recall.
What This Means for AI Regulation
This is one of the most aggressive actions the U.S. government has taken against a major AI company. The order to shut down two flagship models signals that regulators are willing to act quickly when they believe national security is at risk.
For Anthropic, which has built its reputation on being a safety-first AI company, this situation is particularly awkward. The company's own safety warnings and testing may have alerted regulators to a vulnerability, which then led to the shutdown order. In effect, the company's transparency about potential risks may have triggered the very action it now opposes.
Our Take: Safety Transparency Meets Regulatory Reality
This is a complicated situation. Anthropic has always positioned itself as the responsible AI company — the one that takes safety seriously while others rush to market. But this case shows that being transparent about risks can have unintended consequences.
In our view, the government's decision raises important questions. Shutting down models used by hundreds of millions of people is a drastic step. If every narrow security finding leads to a shutdown, AI companies may stop being honest about vulnerabilities. That would make everyone less safe in the long run.
At the same time, national security concerns cannot be ignored. The key question is whether the government acted proportionally. Based on what we know, this appears to be a significant escalation in AI regulation — and one that could change how companies approach safety disclosures going forward.
