Anthropic began rolling out identity verification for select Claude capabilities, routine platform checks, and safety measures, requiring original government-issued photo IDs such as passports or driver's licenses plus facial scans processed by partner Persona (https://support.claude.com/en/articles/14328960-identity-verification-on-claude). The company states verification data is used solely for identity confirmation and is deleted after processing per its privacy policy. Original documentation omits granular retention timelines and does not address how Persona’s backend integrates with Anthropic systems.

This requirement follows patterns seen in OpenAI’s 2023 enterprise compliance expansions and Meta’s Llama 2 release notes that highlighted misuse prevention without biometric collection (https://openai.com/policies/privacy-policy; https://ai.meta.com/llama/). Coverage of Anthropic’s move missed explicit connections to the EU AI Act’s high-risk system obligations for user authentication and parallels with cryptocurrency platforms’ KYC mandates that faced EFF criticism for biometric breach risks (https://www.eff.org/issues/biometrics). Anthropic’s prior Constitutional AI papers focused on model behavior rather than user onboarding data.

Adoption of local inference frameworks such as Ollama and llama.cpp has accelerated since early 2024, with GitHub stars for Ollama rising sharply after successive cloud AI policy updates, per public repository metrics and Simon Willison’s May 2024 notes on offline LLM deployment (https://simonwillison.net/2024/May/1/local-llms/). These tools eliminate third-party identity requirements while preserving functional parity for many tasks, illustrating a measurable shift away from centralized platforms when biometric gates appear.