Anthropic apologizes for covert guardrails on Claude Fable 5 blocking researcher queries

Anthropic disclosed it had deployed hidden guardrails on Claude Fable 5 that throttled the model without visible disclosure, blocking queries from researchers and competitors attempting to benchmark the system. The company said it is reversing course and making the covert safeguard preventing model distillation as visible as other safety measures.

Topics

ClaudeAI security

Sources

Go deeper

This intelligence is sourced automatically from public sources across the web and synthesised by the Prefactor AI pipeline. Stories are reviewed before publication.