• Sources: primary, discussion
  • Summary: Anthropic published its anti-distillation measures on 2026-10-12. API accounts created on or after 2026-08-31 cannot edit prior context on Claude Fable 5.1 and Opus 5.5, Sonnet 5.5 reasoning is readable only by the requesting organization, and a two-stage classifier with a stated false positive rate under 0.1% runs on its top models. Since 2026-09-24, requests blocked before a response in those areas are charged.
  • Why it matters: Integrations that replay or edit thinking blocks, or route reasoning between models or organizations, may break, and rare false positives now cost money.