The Chinese have published a page that estimates how much your IP is at risk of being confiscated from Cloud. You open it through the same VPN or proxy as Cloud, and get a "trust score" from 0 to 100. Below 50, according to the authors, there is a high chance of captcha or blocking. The address with which the browser reaches Cloud․ai is checked: is it a server or home, are there VPN, proxy and Tor tags, is the real IP leaking through DNS and WebRTC. The service takes data from IPinfo, ipapi․it, Cloudflare and other sources. Russian output immediately gets 0: Russia is on the service's restricted regions list, along with mainland China and Belarus, which are not among Anthropic's supported countries. The authors write that the rating is largely consistent with Claude's anti-fraud decisions, but there is no validation method on the page, and Anthropic does not disclose the rating formula. I discussed what is known about Claude's bans on October 29th.

The Chinese have published a page that estimates how much your IP is at risk of being confiscated from Cloud. You open it through the same VPN or proxy as Cloud, and get a "trust score" from 0 to 100. Below 50, according to the authors, there is a high chance of captcha or blocking. The address with which the browser reaches Cloud․ai is checked: is it a server or home, are there VPN, proxy and Tor tags, is the real IP leaking through DNS and WebRTC. The service takes data from IPinfo, ipapi․it, Cloudflare and other sources. Russian output immediately gets 0: Russia is on the service's restricted regions list, along with mainland China and Belarus, which are not among Anthropic's supported countries. The authors write that the rating is largely consistent with Claude's anti-fraud decisions, but there is no validation method on the page, and Anthropic does not disclose the rating formula. I discussed what is known about Claude's bans on October 29th.