Update: Anthropic has walked back this policy after outrage from developers. The company now says Fable 5's safeguards for frontier LLM development will be visible to users instead of silently degrading the model.

I didn't expect to read this in a model card. Fable 5 model card:

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).

Claude can now be silently nerfed. Anthropic has decided it won't tell users when this happens.

Modern software companies increasingly build their own embedding, reranking, and recommendation systems. Even my small bootstrapped app, wanderfugl.com, has a custom reranker and embedding algorithm that I trained myself.

Anthropic gives a few examples of what it considers "frontier AI development," but doesn’t provide a clear line. The problem is that many techniques once reserved for AI labs are now being used by ordinary software companies. Startups train embedding models. They build rerankers. They finetune and host small llms. The boundary between "frontier AI research" and normal product development is becoming harder to define every year.

That creates a real supply chain risk for businesses. If Claude gives me poor or incorrect advice while I’m working on an AI component, I have no way of knowing whether the model was confused, whether my problem is unsolvable, or if some invisible policy restriction quietly kicked in. Anthropic has explicitly chosen not to tell users when this is happening.

Once a development tool can stop optimizing for your success without telling you, it becomes impossible to fully trust your infrastructure.

The Anthropic supply chain risk

Anthropic says these safeguards only affect 0.03% of developers. Maybe that's true today.

The problem is that the definition of an AI company is changing.

Maybe you're not training frontier models today—most companies aren't. But modern software increasingly contains AI models. Five years ago, building a startup meant writing APIs and SQL queries. Today, it often means training, tuning, and deploying models.

Five years ago, models like CLIP were frontier AI research projects. Today I'm fine-tuning them for a bootstrapped travel startup.

If you're debugging a model training pipeline for your product and Claude gives a bad answer, was the model confused? Did you give it bad context? Or did a hidden policy nerf Claude's ability to assist you?

You won't know.