Anthropic has just announced a major privacy policy update that will require Claude users to verify their identities with a government ID and a selfie, and will make it easier for the company to hand over user data to law enforcement without a subpoena. The changes go into effect on July 8, 2026, and signal the end of the anonymous era for large language model services.
Under the new terms, Anthropic may ask free, Pro, and Max Claude users to complete identity verification if it suspects abuse, policy violations, or underage use. The process will be handled by a third-party compliance service called Persona, which will require users to upload a government-issued ID and take a live selfie for comparison. Anthropic says the biometric and ID data won't be stored on its own servers and won't be used for training or marketing.
But the more significant change is in the data-sharing section. The old policy said Anthropic would only disclose user data when "legally required" to do so. The new version lets the company proactively share conversations and other data with law enforcement or government agencies if it has a "good faith belief" that disclosure is "reasonably necessary" to prevent physical harm, fraud, illegal activity, or to enforce its terms of service. This update only applies to consumer accounts — enterprise and API customers remain under the previous, stricter confidentiality terms.
Earlier this month, on June 9, Anthropic also implemented a mandatory 30-day data retention policy for its most powerful models like Fable 5 and Mythos 5, specifically to investigate jailbreaking attempts. While identity verification and proactive data sharing are already standard practice at companies like OpenAI and Google, Anthropic's recent moves come under a cloud of export restrictions and account bans for Fable 5. The consecutive security tightening is prompting privacy-conscious users and businesses to reconsider Claude's privacy boundaries.