Anthropic is significantly updating the privacy policy for its services: starting July 8, 2026, Claude personal account users may be required to undergo identity verification, and the requirements for disclosing data to authorities are becoming less stringent.
What Happened
Under the new rules, Anthropic is implementing the Persona service for identity verification through the provision of ID photos and selfies. This measure will be applied when there are suspicions of violations of the terms of use or when the service is used by minors. Additionally, the company is changing the regulations for transferring data to government agencies: instead of requiring a mandatory court order, a "good faith belief" standard may now be applied.
Context
These changes are occurring against the backdrop of an industry movement toward the standardization of agentic systems and the expansion of AI capabilities. Parallel to the tightening of identity controls at Anthropic, Google Cloud is introducing the Open Knowledge Format (OKF) to simplify knowledge exchange between agents, and Zhipu AI is releasing the GLM-5.2 model with a 1 million token context window.
Why It Matters for the Industry
Anthropic's policy change sets a precedent for shifting the issue of data transfer from a purely legal domain into the realm of the company's internal compliance, which could change privacy approaches in the AI sector. At the same time, the development of open standards, such as Google's OKF, and the growth of context windows among competitors (Zhipu AI) indicate the industry's drive toward creating portable and high-performance agentic ecosystems.
Why It Matters for Users
Claude users should be prepared for the possibility of having to confirm their identity when engaging in suspicious activity. For AI solution developers, the emergence of new tools for structuring knowledge (OKF) and the ability to work with extremely long contexts in new models serve as important signals.
What Is Not Yet Known / Limitations
It is not specified which exact criteria will be considered "suspicious activity" to initiate verification via Persona, nor are the specific procedures for implementing the "good faith belief" principle detailed.
Sources
Author
Look at AI, Editorial Team