Anthropic’s safety warnings may have just backfired — the government has pulled the plug on its most powerful AI
Summary
The US government ordered a recall or shutdown of Anthropic’s most capable deployed model after identifying a potential jailbreak, prompting Anthropic to publicly dispute the decision. Anthropic argues the issue was narrow and not grounds to pull a commercial system used at scale.
Why it matters
Government now has a practical off switch for widely used AI systems, raising the cost of shipping frontier capability without airtight security guarantees.