Hacker News Relitigates Anthropic's Ban on Training Models From Your Own Claude Outputs
A thread asking 'If I own Claude's outputs why can't I train my own model on them?' drew 72 points and 65 comments on the morning of August 13, pointing at Anthropic's support doc: the Terms 'do not allow the use of Outputs to train models that are competitive with Anthropic's own,' explicitly covering general-purpose chatbots, open-ended text generation, using outputs as training targets, and reverse engineering training methods. Anthropic's stated rationale is that customers would otherwise be 'using our infrastructure and investment to build direct competitors.' The policy page itself was last updated March 16, 2026 — this is a community argument about existing terms, not a new restriction, but it's the tension every builder distilling from a frontier model eventually hits.
↳ Follow the thread