In April 2026, Anthropic started requiring government ID verification through Persona before allowing Claude subscriptions. You now need to upload a photo ID to use Claude Pro, Team, or Enterprise plans.
The reaction was immediate and negative. A viral post on X (3,900+ likes) framed it as handing competitors a gift — ChatGPT and Gemini don’t require ID verification.
What changed
Before: sign up with email, add payment method, start using Claude.
Now: sign up with email, verify your identity through Persona (upload government-issued photo ID), add payment method, start using Claude.
Persona is a third-party identity verification service. They process your ID, verify it’s real, and confirm your identity to Anthropic. Persona’s privacy policy governs how your ID data is handled separately from Anthropic’s.
Why Anthropic did this
Anthropic hasn’t given a detailed public explanation, but the timing aligns with two events:
-
Claude Mythos — Anthropic’s model that finds zero-day vulnerabilities. Restricting access to verified identities makes it harder for bad actors to use Claude for malicious purposes.
-
Regulatory pressure — As AI models become more capable, regulators are pushing for “know your customer” requirements similar to financial services.
The practical effect: every Claude user is now tied to a real identity. Anonymous usage of Claude’s paid tiers is no longer possible.
What this means for developers
Your Claude usage is now identity-linked
Every conversation, every API call, every Claude Code session is tied to your verified identity. Combined with the Heppner ruling (AI chats can be subpoenaed), this means your Claude usage is both identifiable and discoverable.
API access still works without ID (for now)
The ID requirement applies to claude.ai subscriptions. API access through the Anthropic API may have different requirements depending on your account type. Enterprise customers negotiate terms directly.
Alternatives that don’t require ID
| Provider | ID required | Quality | Cost |
|---|---|---|---|
| OpenAI (ChatGPT/API) | No | Frontier | $20/mo or API |
| Google (Gemini) | No | Frontier | Free or API |
| DeepSeek | No | Strong | Cheap API |
| Qwen 3.6 Plus | No | Strong | Free API |
| Ollama (local) | No | Good | Free |
| OpenRouter | No | Access to all models | Pay-per-use |
If ID verification is a dealbreaker, OpenRouter gives you access to Claude models through their API without Anthropic’s direct verification requirements (though OpenRouter has its own account requirements).
The privacy trade-off
Anthropic is making a calculated bet: the security benefits of verified users outweigh the users they’ll lose over privacy concerns. For a company that built a model capable of finding zero-day vulnerabilities in every major operating system, this makes sense from a safety perspective.
For developers, the question is whether you’re comfortable with:
- A third party (Persona) processing your government ID
- Your AI usage being tied to your real identity
- The potential for that data to be subpoenaed
If the answer is no, self-hosted models are the only option that gives you both capability and anonymity. See our guides on running models locally and self-hosted AI for enterprise.
Related: Can Your AI Conversations Be Subpoenaed? · AI Data Retention Policies · Best AI Coding Agents for Privacy · OpenRouter Complete Guide · How to Use Claude Code · Self-Hosted AI for Enterprise