TokenMix

AI API, API Gateway, Developer Tools, AI Infrastructure

Visit Website
April 16, 2026 Anthropic now requires government ID for some Claude features — what this means for indie devs building on AI APIs

Hey IH 👋

If you're building anything on Claude (or considering it), heads up: Anthropic rolled out identity verification yesterday. Some Claude users are now being asked for a government-issued photo ID + live selfie before they can access certain features.

The details

  • Physical passport, driver's license, or national ID required

  • Live selfie via camera

  • Verified by Persona (third-party), not Anthropic

  • Triggers are vaguely defined: "certain capabilities," "routine integrity checks," "safety measures"

  • Nobody knows exactly which features trigger it

Why this matters for indie hackers

If you're a solo dev or small team building on AI APIs, provider lock-in risk just got more real. Today it's ID verification. Tomorrow it could be mandatory enterprise contracts for API access.

A few takeaways:

1. API access is unaffected (for now)

This only hits claude web and Claude Code CLI users. If you're calling Claude via API keys, you're fine. But the trend is worth watching.

2. Multi-provider setups are looking smarter

I've been using a unified API gateway for my projects — one endpoint, one API key, swap models by changing a parameter. When one provider adds friction or changes terms, I just route to another. Zero code changes.

If you're hardcoded to a single AI provider, now's a good time to abstract that layer.

3. The indie dev advantage

Big companies negotiate enterprise deals. We don't have that leverage. What we have is agility — we can switch providers in an afternoon. Build with that flexibility in mind.

What I'm doing about it

For my AI products, I access Claude (and GPT, Gemini, DeepSeek, etc.) through a multi-model API gateway rather than direct integration. This means:

  • No individual account-level restrictions affect my production apps

  • If pricing or policies change at one provider, I can switch in minutes

  • One bill, one API key, 150+ models

Discussion

How are you handling AI provider dependency in your stack? Are you locked into one provider or do you have a multi-model setup?

Curious to hear how other IH builders are thinking about this.

About

I kept switching between AI providers — OpenAI one week, Claude the next, then DeepSeek when pricing changed. Every switch meant a new SDK, new API key, new billing account, new error handling.