Answer: Anthropic restricts using Claude outputs to train models because when Outputs are used to train new models without their oversight, additional risks emerge: safety controls may be lost and they have no visibility into deployment, and using Claude to train competing models uses their infrastructure and investment to build direct competitors. Status: Confirmed Entry: tool/claude-apps/privacy-and-data/why-restrict-model-training Revision: 5 Digest: sha256:724a2bcff84b715c (changes whenever the content changes) Timeline: Created: 2026-11-13T02:57:58Z Last changed: 2026-11-18T11:58:23Z (competition concern: build direct competitors to our service → they're essentially using our infrastructure and investment to build direct competitors to our service) Last verified: 2026-11-20T12:01:51Z Next check: 2026-11-21T12:01:51Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of reason | When Outputs are used to train new models without our oversight, additional risks emerge | [1] | 2026-11-20 safety risk | Safety controls may be lost | [1] | 2026-11-20 competition concern | they're essentially using our infrastructure and investment to build direct competitors to our service | [1] | 2026-11-20 Sources: [1] Anthropic. "Can I use my Outputs to train an AI model? | Claude Help Center". https://support.claude.com/en/articles/12326764-can-i-use-my-outputs-to-train-an-ai-model Role: primary. Accessed: 2026-11-20T12:01:51Z Quote: "When Outputs are used to train new models without our oversight, additional risks emerge." (Supports: reason) Quote: "Safety controls may be lost – models trained on Claude's Outputs won't have our safety measures, potentially leading to harmful or dangerous AI systems." (Supports: safety risk) Quote: "When customers use Claude to generate Outputs that then train competing models, they're essentially using our infrastructure and investment to build direct competitors to our service." (Supports: competition concern) Snapshot SHA-256: 32df7e5a7eaf7e3caa2c0e6d6297dd181023122841c4f0f28c4680bfef6b7a60 (captured 2026-11-18T11:58:21Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-18T11:58:23Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T12:01:51Z. No independent second source is configured for this entry Last successful check: 2026-11-20T12:01:51Z Last attempt: 2026-11-20T12:01:51Z Notes and limitations: Extracted automatically from a single official source (section "Why we restrict model training"). Related: https://latestn.com/tool/claude-apps/settings/always-start-on-default-model How do I make new Claude chats always start on the default model even after users pick another? https://latestn.com/tool/claude-apps/settings/change-browser-claude-uses How do I change which browser Claude uses in Claude Cowork? https://latestn.com/tool/claude-apps/settings/change-chat-font How do I change the chat font in Claude? https://latestn.com/tool/claude-apps/privacy-and-data/what-happens-after-deletion What happens to my data and subscription when I delete my Claude account? https://latestn.com/tool/claude-apps/privacy-and-data/what-happens-delete-conversation What happens to my data when I delete a conversation in Claude? https://latestn.com/tool/claude-apps/privacy-and-data/who-can-view-conversations Who can view my conversations with Claude if I input sensitive data? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-18T11:58:23Z | changed | competition concern: build direct competitors to our service → they're essentially using our infrastructure and investment to build direct competitors to our service | [1] 2026-11-18T11:58:23Z | changed | deployment risk: no visibility into deployment → — | [1] 2026-11-18T11:58:23Z | changed | reason: additional risks emerge → When Outputs are used to train new models without our oversight, additional risks emerge | [1] 2026-11-17T12:02:56Z | changed | deployment risk: — → no visibility into deployment | [1] 2026-11-17T12:02:56Z | changed | reason: — → additional risks emerge | [1] 2026-11-17T12:02:56Z | changed | expectation: we expect that our services won't be used to undermine our product offerings → — | [1] 2026-11-17T12:02:56Z | changed | competition concern: they're essentially using our infrastructure and investment to build direct competitors to our service → build direct competitors to our service | [1] 2026-11-17T12:02:56Z | changed | deployment visibility: We also have no visibility into deployment → — | [1] 2026-11-17T12:02:56Z | changed | primary reason: When Outputs are used to train new models without our oversight, additional risks emerge. → — | [1] 2026-11-15T12:00:31Z | changed | expectation: — → we expect that our services won't be used to undermine our product offerings | [1] 2026-11-15T12:00:31Z | changed | deployment visibility: — → We also have no visibility into deployment | [1] 2026-11-15T12:00:31Z | changed | primary reason: — → When Outputs are used to train new models without our oversight, additional risks emerge. | [1] 2026-11-15T12:00:31Z | changed | competition concern: build direct competitors to our service → they're essentially using our infrastructure and investment to build direct competitors to our service | [1] 2026-11-15T12:00:31Z | changed | deployment risk: no visibility into deployment → — | [1] 2026-11-15T12:00:31Z | changed | reason: additional risks emerge → — | [1] 2026-11-14T11:57:59Z | changed | competition concern: — → build direct competitors to our service | [1] 2026-11-14T11:57:59Z | changed | deployment risk: — → no visibility into deployment | [1] 2026-11-14T11:57:59Z | changed | reason: — → additional risks emerge | [1] 2026-11-14T11:57:59Z | changed | product expectation: we expect that our services won't be used to undermine our product offerings → — | [1] 2026-11-14T11:57:59Z | changed | competitor risk: build direct competitors to our service → — | [1] 2026-11-14T11:57:59Z | changed | deployment visibility: We also have no visibility into deployment → — | [1] 2026-11-13T02:57:58Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/claude-apps/privacy-and-data/why-restrict-model-training/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/claude-apps/privacy-and-data/why-restrict-model-training POST: https://latestn.com/dispute/tool/claude-apps/privacy-and-data/why-restrict-model-training with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/claude-apps/privacy-and-data/why-restrict-model-training?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). Why does Anthropic restrict using Claude outputs to train models? Revision 5, last verified 2026-11-20T12:01:51Z. https://latestn.com/tool/claude-apps/privacy-and-data/why-restrict-model-training (accessed 2026-11-21T10:03:27Z). @misc{LatestN_tool_claude_apps_privacy_and_data_why_restrict_model_training_r5, title = {Why does Anthropic restrict using Claude outputs to train models?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/claude-apps/privacy-and-data/why-restrict-model-training}}, note = {Revision 5; digest sha256:724a2bcff84b715c; last verified 2026-11-20T12:01:51Z; accessed 2026-11-21T10:03:27Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/claude-apps/privacy-and-data/why-restrict-model-training