Read the Anthropic blog post here.

Anthropic will update its Privacy Policy, effective July 8, 2026, to expand the types of user data it collects and how it uses that data. These changes apply only to consumer accounts, including Claude Free, Pro, and Max plans. They do not affect Claude Team, Enterprise, the Developer Platform, or services governed by separate commercial agreements.

As Claude handles more complex, multi-step tasks and integrates with third-party apps and services, Anthropic now discloses what data is shared with external services and what data it receives in return after task completion.

Verification may require a government ID and a selfie: According to Anthropic’s support page, Claude users may encounter verification prompts when using specific features, during platform integrity checks, or for safety and compliance purposes.

The company states that users may need a government-issued photo ID and a camera-equipped device. Accepted IDs include passports, driver’s licences, state or provincial ID cards, and national identity cards. The ID must be original, physical, government-issued, legible, undamaged, and include a photograph. Anthropic does not accept photocopies, screenshots, scans, photos of photos, digital or mobile IDs, student IDs, employee badges, library cards, bank cards, or temporary paper IDs.

Persona handles ID and selfie collection: Anthropic uses Persona Identities as its identity verification partner. According to Anthropic’s support page, Anthropic acts as the data controller for verification data, while Persona processes that data on its behalf. ID and selfie images are collected and stored by Persona, not on Anthropic’s systems.

Anthropic can access verification records through Persona’s platform during appeals. Persona is contractually limited to using the data for verification, support, and fraud prevention. The data is encrypted both in transit and at rest and is not used to train models.

Persona must delete verification data in accordance with Anthropic’s retention limits and applicable law, but the support page does not specify a retention period. Since verification requires government IDs and live selfies, this omission is significant, especially for non-US users who may be uncertain about where their data is stored, who can access it, and which jurisdiction applies.

Verification failures can affect account access: Anthropic states that verification may fail because of a blurry photo, an unreadable document, an expired ID, or technical issues. Users are advised to retry with better lighting, ensure that the document is clear and valid, or contact Anthropic if problems persist.

The company notes that accounts may be banned after verification for repeated policy violations, account creation from unsupported locations, violations of the Terms of Service, or use by individuals under 18. Users who believe their account was banned in error may appeal through Anthropic’s online form.

Unanswered questions persist over Anthropic’s verification practices: MediaNama previously reported that Reddit and X users, including Claude Pro subscribers, claimed that Anthropic wrongly flagged them as minors and suspended their accounts, citing detected signals of underage use and providing a 30-day window for age verification. Some accounts were reinstated after ID verification, while others experienced ongoing project disruptions despite receiving refunds. Anthropic has said that checks may be triggered by certain capabilities, platform integrity reviews, or safety and compliance reasons, without specifying which features or behaviours lead to such reviews.

MediaNama raised questions about the criteria behind these flags, how users can appeal, and how the process aligns with India’s Digital Personal Data Protection Act, 2023, particularly regarding the transfer of verification data to Persona’s US servers, the legal basis for that transfer, and applicable retention periods.

It also highlighted concerns about Persona itself, citing researchers’ findings on watchlist screening, adverse media checks, risk scoring, and multi-year data retention in Persona’s files. While there is no evidence that verification data is being shared with government systems, the report argued that the partnership sits awkwardly alongside Anthropic’s public stance against surveillance-enabling AI uses.

Model training and verification data are separate issues: Anthropic says verification data is not used to train its models. This policy should be distinguished from Claude’s broader model-training approach. According to Anthropic’s privacy policy, user inputs and outputs may be used to improve services and train models unless users opt out through their account settings. The policy also states that conversations may be used to improve the model if the content is flagged for safety review or explicitly reported through user feedback.

Also read:

• Anthropic calls for the ability to slow or pause frontier AI development
• Anthropic To Utilise User Chats For AI Training: What Are Its Risks?
• Anthropic opens Mythos-class AI to the public, but keeps high-risk capabilities behind guardrails