If I own Claude's outputs why can't I train my own model on them?
support.claude.com
August 13, 2026
2 min read
🔥🔥🔥🔥🔥
46/100
Summary
Users retain ownership of Outputs generated by Claude from their Inputs. Training or developing AI models using these Outputs is prohibited without written permission from Claude.
Key Takeaways
- Users own the Outputs generated from their Inputs when using Claude, but cannot use these Outputs to train AI models without written permission from Anthropic.
- Anthropic prohibits the use of Claude's Outputs to develop competing models to protect their safety measures and prevent misuse.
- Allowed uses of Outputs include training non-competing models for tasks like sentiment analysis, content categorization, and anomaly detection.
- Prohibited uses include training general-purpose chatbots and models for open-ended text generation, as well as reverse engineering training methods.
Community Sentiment
Positives
- Some commenters see the current ownership definitions as a potential catalyst for innovation in AI, suggesting that clearer rules could spur new models and competitors.
Concerns
- The definition of ownership is seen as misleading, with many commentators feeling that users are misled into thinking they own outputs when they actually only have a license.
- Concerns about copyright and licensing dominate the discussion, with fears that AI companies are exploiting open-source content without proper credit or respect for licenses.
- The sentiment that AI companies have taken data without permission raises significant ethical concerns, with users feeling their contributions are being unfairly appropriated.
