Anthropic just pulled off its hardest move of 2026: getting a banned AI model reinstated by the US government in 19 days. On June 12, Amazon engineers reported to the Commerce Department that Claude Fable 5 could identify software vulnerabilities and produce exploitable code. Hours later, both Fable 5 and Mythos 5 went dark globally. On July 1, they came back — the result of two weeks of Washington negotiations, a new safety classifier, and an industry jailbreak framework...