Anthropic moves to close loopholes that allow Chinese access to Claude
Anthropic just discovered engineers are still routing Claude through proxies. Here's why your export controls might already be obsolete.
Why it matters
As AI models become geopolitical assets, companies face a cat-and-mouse game with access restrictions. Anthropic's enforcement gap reveals how technical controls lag behind determined circumvention—a pattern that will shape AI regulation and corporate compliance strategies globally.
The key facts
4 to knowAnthropic identifying and closing loopholes enabling Chinese access to Claude
Engineers finding workarounds despite stringent restrictions
Highlights ongoing tension between AI accessibility and export/sanctions compliance
Suggests technical enforcement of geo-restrictions remains a governance challenge
Go to the source
Financial Times Technologyft.com
Publisher excerpt: Engineers are still finding ways to use AI models despite stringent restrictions