OpenAI
1970-01-01
OpenAI and Anthropic Mandate Identity Verification: End of Anonymous AI Era
Summary
Key Takeaways
On June 17, 2026, ChatGPT users received pop-ups demanding ID upload and 3D liveness detection video. OpenAI had quietly deployed an age prediction model in January 2025, using four signals (e.g., account age, usage patterns) to silently flag minors. In April 2025, OpenAI launched Verify Organization for developers, requiring government-issued entity IDs, each valid for only one organization per 90 days. The latest help center page shows verification expanding from flagged users to general users.
A week earlier, on June 10, 2025, Anthropic notified users of a privacy policy update effective July 8, introducing identity and age verification. The policy states Claude users may be verified when using specific features, during platform integrity checks, or for safety/compliance. Anthropic's email stressed: AI is evolving from a chat tool to an agent executing real actions (booking flights, editing files, operating third-party apps). Once AI acts on your behalf, who gives the command becomes a legal issue, not just a security one.
Why It Matters
Beneath the safety veneer lies a coordinated move to build a user identity database, permanently binding behavior to real identities. This raises privacy cost traps: ID photos and 3D liveness videos are highly sensitive biometric data, stored on third-party clouds (AWS/Azure) with leak risks. The verification process introduces tail latency that disrupts real-time AI interactions, especially in high-frequency coding assistant scenarios.
This is a control plane shift: the identity verification system becomes a new control point, allowing platforms to deny service arbitrarily for compliance reasons, stripping user architectural flexibility. It effectively encircles open-source model camps (Llama, Mistral), which cannot enforce identity verification, pushing enterprises toward closed-source platforms for compliance.
PRO Decision
【Vendors】Competitors (Google DeepMind, Meta AI, Mistral AI) should launch anonymous usage + agent identity separation alternatives: allow anonymous basic chat, trigger temporary verification only for real-world actions (payments, orders), avoiding permanent binding. Lobby regulators for unified identity standards to prevent OpenAI/Anthropic from creating a de facto verification monopoly.
【Enterprises】CIOs and architects must conduct zero-trust technical audits: assess verification impact on internal AI workflows, demand data localization options for biometric data. Establish verification failure fallback mechanisms to prevent business disruption. Compare open-source models under compliance frameworks to avoid lock-in by closed platforms.
【Investors】See through the PR: identity verification is a user stickiness enhancer but also increases churn risk (privacy-conscious users flee to anonymous alternatives). Short-term positive for OpenAI/Anthropic's compliance moat; long-term watch for user growth slowdown and regulatory backlash (EU strict rules on biometric storage). Consider reducing exposure to verification-dependent assets, increase holdings in privacy-preserving + anonymous AI startups.
Get 3-5 key AI infrastructure signals weekly →
💬 Comments (0)