- Anthropic launched Claude Fable 5 and Mythos 5 — but built in secret safeguards that can quietly downgrade or reroute answers if the model suspects a user is doing frontier AI research.
- Critics say those limits blur safety and competitive strategy, potentially hobbling outside researchers and concentrating power among top labs.
- Reactions are split: some call it responsible caution, others call it marketing or anti‑competitive — and the models’ strong capabilities and pricing have intensified the debate.
The news, distilled into what matters.
Get the full experience in the app — topics, comments, and audio summaries.