By default, your conversations with Claude can be used to train Anthropic's AI. You can opt out in account settings, but any conversation you rate (thumbs up or down) or that gets flagged for safety reasons can still be used for training even if you opted out.
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
The opt-out does not provide a complete exclusion from model training; two specific categories of conversations remain eligible for training use regardless of the opt-out setting, which affects the practical scope of the privacy control offered.
Consumers who opt out of model training should be aware that conversations they rate using the feedback interface, or conversations flagged by Anthropic's safety systems, remain available for training use under these terms. The opt-out applies only to conversations that do not fall into these two categories.
How other platforms handle this
We may use the content you provide to us, including prompts and generated images, to train and improve our AI models and services.
We may leverage OpenAI models independent of user selection for processing other tasks (e.g. for summarization). We may leverage Anthropic models independent of user selection for processing other tasks (e.g. for summarization). We may leverage these models independent of user selection for processi...
Users under 18 years old interact with an age-appropriate model specifically designed to reduce the likelihood of exposure to sensitive or suggestive content. Our under-18 model has additional and more conservative classifiers than the model for our adult users so we can enforce our content policies...
Monitoring
Anthropic has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 10 platforms.
"We may use Materials to provide, maintain, and improve the Services and to develop other products and services, including training our models, unless you opt out of training through your account settings. Even if you opt out, we will use Materials for model training when: (1) you provide Feedback to us regarding any Materials, or (2) your Materials are flagged for safety review to improve our ability to detect harmful content, enforce our policies, or advance our safety research.— Excerpt from Anthropic's Anthropic Consumer Terms
(1) REGULATORY LANDSCAPE: This provision implicates GDPR Articles 6 and 9 (legal basis for processing personal data for model training), Article 7 (conditions for consent), and Recital 32 regarding granularity of consent. The carve-out for safety-flagged content and feedback raises questions about whether the opt-out mechanism constitutes meaningful consent withdrawal under GDPR, which the relevant supervisory authorities (EU data protection authorities, UK ICO) may evaluate. CCPA is also relevant regarding whether model training constitutes a 'sale' or 'sharing' of personal information. The EU AI Act may engage requirements around training data transparency for general-purpose AI models. (2) GOVERNANCE EXPOSURE: Medium-High. The feedback and safety review carve-outs mean that the opt-out mechanism does not align with a full right to object to processing under GDPR Article 21, creating potential tension with EU and UK data protection frameworks. The broad framing of 'safety review' is not defined with specificity in the document, leaving the scope of that exception uncertain. (3) JURISDICTION FLAGS: EU and UK users face heightened exposure because GDPR and UK GDPR impose specific requirements on the legal basis for processing personal data for AI training purposes, and the adequacy of an opt-out (versus affirmative consent or legitimate interests assessment) is actively scrutinized by European DPAs. California residents may have rights under CCPA to know and limit use of personal information. The provision's application in jurisdictions with sector-specific AI regulations warrants monitoring. (4) CONTRACT AND VENDOR IMPLICATIONS: Organizations using Claude.ai under corporate accounts and relying on this consumer ToS should note that the training opt-out exceptions may not satisfy GDPR Article 28 data processing agreement requirements if employee data is processed. Procurement teams should evaluate whether a Data Processing Addendum is available and whether the safety flagging exception is consistent with data minimization obligations. (5) COMPLIANCE CONSIDERATIONS: Compliance teams should document the scope of the opt-out limitation in their records of processing activities. If downstream products are built on Claude.ai, privacy notices should accurately reflect that feedback-related and safety-flagged interactions may be used for training. Legal teams in EU/UK jurisdictions should assess whether the current legal basis for training on flagged content is clearly established in Anthropic's privacy documentation.
Full compliance analysis
Regulatory citations, enforcement risk, and due diligence action items.
Free: track 1 platform + weekly digest. Watcher: 10 platforms + same-day alerts. No credit card required.
How 10 AI platforms describe the use of user data for model training, improvement, and development, based on archived governance provisions.
Professional Governance Intelligence
Need to monitor specific governance provisions?
Professional includes provision-level monitoring, governance timelines, regulatory mapping, and audit-ready analysis.
Built from archived source documents, structured governance mappings, and historical version tracking.
The opt-out does not provide a complete exclusion from model training; two specific categories of conversations remain eligible for training use regardless of the opt-out setting, which affects the practical scope of the privacy control offered.
Consumers who opt out of model training should be aware that conversations they rate using the feedback interface, or conversations flagged by Anthropic's safety systems, remain available for training use under these terms. The opt-out applies only to conversations that do not fall into these two categories.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Anthropic.