Claude's 18+ age check goes live — flagged accounts get locked out
Anthropic's age-assurance system is flagging Claude accounts as under-18 and demanding a Yoti ID scan to get back in. Plus a compiler graph for C# agents.
Copy markdown
The age gate is live — and it locks you out
Anthropic's age-assurance system is actively flagging Claude accounts it suspects belong to under-18s and disabling them until you prove your age. It covers the consumer product — Free, Pro, and Max on claude.ai — and reinstates you only after a Yoti check: a selfie for facial age estimation, a government ID scan, or the Yoti Digital ID app. No pass, no Claude.
Reportedly, a chat review can trip the flag — and you get 30 days
Users on Reddit and X report Pro accounts restricted after Anthropic reportedly reviewed their conversations and judged them minors, with the lockout notice giving 30 days to appeal before access is permanently revoked. Whatever the trigger, the takeaway for builders is blunt: your chats can be read in a moderation review, so treat claude.ai like any logged service.
Graphify C#: compiler-accurate “find usages” for your agent
A new MIT-licensed, headless Roslyn/MSBuild indexer turns a C# solution into a queryable JSON graph of compiler-resolved calls, references, overrides, and interface implementations — the exact symbol resolution grep can't manage across overloads and generics. Run `graphify-csharp --input ./src/App.sln --output out.json` and hand the graph to your coding agent so it stops guessing which overload actually fired.
Elsewhere: GPT Image 2.5 splits into Flare and Sunburst
OpenAI's GPT Image 2.5, out last week, ships two API models: Flare (the default, ~50% lower latency) and Sunburst (tighter multi-turn edit control), both with a sketch-to-image tool, transparent backgrounds, and 4K output. Same token pricing either way — about $0.006 for a low-quality 1024× image up to $0.21 at max quality.