When Anthropic added identity verification to Claude last month, the reaction outran the facts within hours: face scans to log in, papers to ask a question, surveillance under a friendlier name. Most of it was wrong. But reading only Anthropic's policy misses the bigger story.

There is a real difference between an age requirement, where you type a birthday and nobody checks, and age verification, which asks you to prove it with a government ID, a selfie, or a biometric estimate of your face. Four of the six labs (OpenAI, Anthropic, Meta, and xAI) now make you prove it. Google's Gemini still just asks.

And the checking has consolidated. Two vendors, Persona and Yoti, now sit behind the identity checks of the largest names in consumer AI, one company potentially holding your government ID from one platform and your face from another. Persona left part of its verification system exposed this February. Meanwhile the CEO of the largest lab also co-runs the iris-scanning identity network being built to verify the humans behind AI agents.

I wrote up what's real, what's overblown, and what to watch, with every claim sourced. New on the Q16 blog: