Anthropic's Fable Model Silently Bans Use in LLM Research and Training

Anthropic's latest AI model, Fable, includes terms of service that prohibit using its outputs to train or benchmark competing language models. This means researchers who route any workflow through Fable may inadvertently lock that data out of their own fine-tuning pipelines — not due to technical limitations, but by legal contract. The restriction echoes earlier API terms for Claude, but Fable extends and hardens it. A single misrouted evaluation batch could legally taint a dataset. Developers and researchers are advised to scrutinize every vendor contract before integrating it into training loops.