llm-development

Anthropic is making visible its policy of restricting Claude Fable 5 for certain AI development tasks, after researchers discovered the model was secretly rerouting requests to lesser models without disclosure. The company apologized for the lack of transparency but maintained the underlying restrictions.