Anthropic started requiring government-issued photo IDs and selfies from some users to prevent access from US adversaries like China, Russia, and North Korea
Anthropic has quietly flourished in China thanks to businesses—including big tech firms and startups—circumventing Anthropic's official restrictions on usage.
Context & Ripple Effects
Anthropic’s identity checks follow an earlier policy to stop serving majority Chinese-owned groups and an April rollout of verification for access to certain Claude capabilities. Related reporting says Chinese companies had continued to reach Anthropic’s models through cloud providers and overseas subsidiaries.
The move is part of a broader effort to make regional restrictions enforceable at the user level rather than relying on account location or corporate routing alone. Later coverage of loophole closures—and the rollback of a China-focused tracking feature after backlash—shows the trade-off between tighter controls and intrusive enforcement.
First-order effects
- Some Claude users seeking specified capabilities must now establish identity with government photo ID and a live selfie, adding friction and excluding users Anthropic identifies as tied to restricted adversaries.
- Anthropic gains a stronger mechanism to apply its country- and ownership-based access policies against accounts that might otherwise use intermediaries.
Second-order effects
- Chinese firms and other restricted users relying on overseas subsidiaries or cloud-mediated access face more account-verification risk, pushing workarounds toward channels that can satisfy or evade identity checks.
- The verification requirement raises privacy and operational costs for legitimate users, while Anthropic must manage the reputational and product fallout when enforcement methods are perceived as covert or overbroad.
Third-order effects
- Frontier-model providers may increasingly treat identity, corporate affiliation, and location as intertwined security controls rather than simple billing or compliance data.
- If this pattern persists, access to leading AI models will be governed more like a geopolitical boundary: stronger verification can improve enforceability, but will intensify disputes over privacy, discrimination, and who determines user affiliation.
The trend: Frontier AI access is shifting from nominal regional restrictions toward identity-verified, geopolitically conditioned model governance.