Anthropic is stepping up enforcement against Chinese companies and engineers circumventing its ban on access from mainland China, including through Singapore-based corporate accounts and VPN-accessed personal subscriptions. The company uses detection systems to identify and block violating accounts. Chinese engineers have sought access partly because Claude Code's outputs can be used for distillation—a technique to train smaller models to mimic more capable ones.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Anthropic is cracking down on workarounds that Chinese companies have used to access Claude AI, including corporate accounts routed through Singapore entities and personal subscriptions accessed via VPN. The company said it explicitly prohibits access from unsupported regions including China and uses detection systems to identify and ban violating accounts.
Why it matters
Chinese engineers value Claude Code—which produces outputs usable for distillation, a technique to train smaller AI models—making unauthorized access a competitive concern for Anthropic. While the workarounds breach Anthropic's terms of service, they do not violate U.S. or Chinese law, leaving Anthropic to rely on its own enforcement.
What to watch
Anthropic continues to target 'transfer station' services that relay mainland requests through overseas accounts, though larger Chinese AI groups avoid these due to concerns that operators may store or resell prompts.
Anthropic has long maintained one of the strictest access bans among U.S. AI companies toward China, but the company now faces a gap between its policy and enforcement. Chinese enterprises—particularly Ant Financial and ByteDance—have systematically found ways to work around the restriction by routing access through third-country entities and personal subscription schemes. These tactics exploit Anthropic's enforcement surface area: the workarounds are not illegal under U.S. or Chinese law, so Anthropic must rely on its own detection and account termination capabilities rather than legal intervention.
The particular appeal of Claude for Chinese engineers points to a strategic concern beyond simple access: Claude Code's ability to generate outputs suitable for distillation—where smaller, cheaper models are trained to replicate the behavior of more capable ones—makes it valuable for cost-conscious AI development. Larger Chinese AI groups have notably avoided third-party relay services out of fear that operators may store or resell prompts, suggesting they maintain at least some awareness of operational risk even as they seek alternative pathways. Anthropic's reliance on continuously evolving detection systems suggests this has become an ongoing technical and operational challenge rather than a one-time closure.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
