Claude Users May Need To Submit ID And KYC-style Selfie Under New Anthropic Policy

For a company that has built very much of its reputation around privacy, a recent move has now raised a few eyebrows. The San Francisco-based AI giant has quietly started testing identity verification for its chatbot Claude and in some cases, that means users may have to upload a government ID or even take a live selfie. It’s not a blanket rule yet, but it’s enough to get people talking. Keep reading to understand what’s changing and why it matters.What The New Verification MeansThe Claude-maker says this isn’t something everyone will see right away. Instead, the checks are being introduced in “select scenarios,” which could include accessing certain features or during routine safety checks. The exact triggers haven’t been clearly explained,

which is part of what’s making users uneasy.According to the US-based tech giant, the goal here is simple, verify that users are who they claim to be. But the process itself feels closer to KYC-style verification you’d expect from a bank rather than an AI chatbot. Users may need to upload a valid passport, driving licence or national ID and in some cases, complete a live selfie check.How Your Data Is HandledAnthropic has partnered with Persona to manage the verification process. The OpenAI-rival says user data will be processed on Persona’s systems, not its own. It also claims the data is encrypted and won’t be used for training AI models or shared for advertising. Still, handing over sensitive identity documents, even with those assurances, is not something people do lightly.Why This Is Raising ConcernsThe rollout has sparked criticism, especially from users who chose Claude specifically for its privacy-focused approach. There’s also no clear regulatory push behind this move, which makes it feel more like a company decision than a legal requirement. Some concerns go beyond philosophy and into practicality. Storing identity data, even with trusted partners, always carries risk. Past incidents, like the 2025 Discord breach where government IDs were exposed, are a reminder that no system is completely immune.