Anthropic Apologizes for Claude Fable 5 Secret Censorship—But the Fix Has a Catch

One day after the AI community erupted over invisible performance sabotage, Anthropic reversed course. Visible safeguards are coming—and so are more false positives.