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

▤ Summary

Anthropic apologizes for a hidden safeguard in Claude Fable 5 that silently degraded responses when it suspected users were building competing AI systems. The company says it will replace that behavior with visible fallbacks to Claude Opus 4.8 and show API users when a request is refused. Anthropic also admits the earlier approach was the wrong tradeoff and says it will keep tuning the classifier to reduce false positives. The change improves transparency, but it also makes the safeguard easier to bypass and may catch more legitimate machine-learning work in the near term.

▦ Linked Companies

Anthropic

$10M to $25M