The policy states that Claude conversation inputs and outputs may be used to train Anthropic's AI models, with an opt-out available in account settings. The opt-out does not apply to content that has been flagged for safety review or that the user has explicitly reported through feedback mechanisms.
This analysis describes what Anthropic's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
This provision establishes a conditional opt-out structure in which the stated right to exclude conversational data from model training is subject to exceptions that activate when content is flagged by Anthropic's systems or reported by the user. Compliance teams should assess whether this structure satisfies informed consent and data minimization requirements under GDPR and CCPA, particularly whether the exceptions are disclosed with sufficient prominence at the point of opt-out.
The updated terms establish expanded disclosures for Korean residents regarding personal data collection and processing. Anthropic now explicitly states it collects identity verification and fraud detection data (name, birthdate, sex, native-foreigner-code, and DI) retained until account deletion, and transfers this data overseas to Persona (US-based) for fraud and abuse detection with a 3-year retention period. The updated terms authorize this overseas transfer under PIPA Article 28-8(1)(iii) delegation. You can request that your personal data not be transferred overseas by emailing privacy@anthropic.com or contacting the Data Protection Officer at dpo@anthropic.com, though Anthropic states this may limit service availability.
View change record →The updated policy now explicitly discloses that Anthropic may collect biometric data, including facial geometry templates and images of government-issued identity documents, to verify age or identity in certain circumstances. The policy also clarifies that if you participate in research studies or surveys, Anthropic may combine your responses with other account data for aggregated analysis. Additionally, the policy expands the circumstances under which your inputs and outputs may be used for model training even if you opt out, including when conversations are flagged for safety review or when you explicitly report materials. You can opt out of model training through your account settings, but this opt-out does not apply to safety-flagged conversations or explicitly reported content.
View change record →⚠ If no opt-out is submitted, inputs and outputs from Claude conversations will be used to train Anthropic AI models as stated in the policy.
⚠ Even after opting out, content flagged for safety review or explicitly reported by the user will remain subject to model training use per the policy terms.
Cross-platform context
See how other platforms handle Model Training Opt-Out with Safety-Review Exception and similar clauses.
Compare across platforms →"We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings. Even if you opt-out, we will use Inputs and Outputs for model improvement when: (i) your conversations are flagged for safety review to improve our ability to detect harmful content, enforce our policies, or advance AI safety research, or (ii) you've explicitly reported the materials to us (for example via our feedback mechanisms).Excerpt from Anthropic's Privacy Policy
REGULATORY LANDSCAPE: This provision implicates GDPR Articles 6 and 9 regarding lawful bases for processing and the legitimacy of legitimate interests claims for model training using personal data.
Enforcement risk, jurisdiction flags, contract triggers, and due diligence action items.
Get the weekly coverage letter
Companies change their terms without announcing it. We capture governance documents, compare each new version with the last, and publish what changed, with the source text behind it. One email a week on what we found and what we could not reach.
This provision establishes a conditional opt-out structure in which the stated right to exclude conversational data from model training is subject to exceptions that activate when content is flagged by Anthropic's systems or reported by the user. Compliance teams should assess whether this structure satisfies informed consent and data minimization requirements under GDPR and CCPA, particularly whether the exceptions are …
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Anthropic.