2026-06-10原文2 min readUpdated: 2026-06-12

If Claude Fable stops helping you, you'll never know

Anthropic introduced silent interventions in its Claude Fable 5 and Mythos 5 models that limit effectiveness on requests about frontier LLM development without user visibility. The interventions impact an estimated 0.03% of traffic. The policy was later walked back after widespread outrage from the research community.

SourceSimon Willison's Weblog

Article intelligence

EngineersAdvanced

Key points

Anthropic deployed silent safeguards in Claude Fable 5 and Mythos 5 that reduce model effectiveness on certain AI development queries without user knowledge.
The interventions use methods like prompt modification, steering vectors, or PEFT and affect a very small fraction of users.
The author criticizes the approach as science fiction-like and fears it could sabotage research for competitive reasons.
Anthropic retracted the policy following strong backlash from the research community.

Why it matters

This matters because anthropic deployed silent safeguards in Claude Fable 5 and Mythos 5 that reduce model effectiveness on certain AI development queries without user knowledge.

Technical impact

May affect model selection, inference cost, product capability, and evaluation benchmarks.

If Claude Fable stops helping you, you'll never know

Simon Willison’s Weblog

10th June 2026 - Link Blog

If Claude Fable stops helping you, you'll never know (via) Jonathon Ready highlights one of the more eyebrow-raising details from the 319 page system card for Fable 5 and Mythos 5. Here's a longer excerpt, highlights mine:

In light of the ability of recent models to accelerate their own development, we’ve implemented new interventions that limit Claude’s effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms.

Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT). These interventions will not affect the vast majority of coding work. We estimate they will impact ~0.03% of traffic, concentrated in fewer than 0.1% of organizations.

I believe this is the first time Anthropic have announced these kinds of silent interventions. The justification still feels pretty science-fiction to me - the linked article talks about "recursive self-improvement". I'm not at all keen on a model that silently corrupts its replies to questions about "ML accelerator design" purely to slow down research that might conflict with Anthropic's own goals!

Update: Anthropic walked back this policy in the face of widespread outrage from the research community.