Anthropic is strengthening its efforts to prevent unauthorized access to its artificial intelligence services from China, according to a report by the Financial Times.
The company has reportedly identified several methods used by Chinese businesses to bypass restrictions on access to Claude, its flagship AI platform.
Chinese Companies Used Overseas Workarounds
The Financial Times reported that Chinese companies, including Ant Financial, accessed Claude through overseas subsidiaries, cloud providers and other indirect methods.
Ant reportedly gave employees corporate Claude accounts connected to its Singapore-based business entity.
ByteDance, meanwhile, is said to have reimbursed engineers for personal Claude subscriptions accessed through virtual private networks, or VPNs.
Access Methods Breached Anthropic’s Rules
The reported practices do not appear to violate U.S. or Chinese law.
However, they allegedly breach Anthropic’s terms of service.
The company’s rules prohibit Chinese businesses, as well as foreign entities controlled by them, from using Anthropic’s AI models.
Anthropic Expands Account Monitoring
Anthropic has reportedly increased its efforts to detect and close unauthorized accounts.
According to the Financial Times, the company is monitoring account activity for signs that users may be located in restricted regions.
These checks may include examining computer time zones and other account signals.
Transfer Services Also Face Scrutiny
Anthropic is also targeting so-called transfer station services.
These platforms allow users to send requests through overseas Claude accounts, creating an indirect route to the AI service.
By closing these access points, Anthropic aims to make it more difficult for restricted companies to use Claude through third parties.
Cloud Platforms Used to Access Claude
Some Chinese businesses reportedly accessed Claude through foreign subsidiaries using international cloud infrastructure.
The Financial Times said Microsoft Azure was among the cloud platforms used in some cases.
This suggests that Anthropic’s enforcement efforts may need to cover both direct accounts and access provided through external cloud services.
Anthropic Tightens Control Over Claude Access
The latest measures highlight the growing challenge facing AI companies as they enforce regional restrictions across global cloud networks.
Anthropic is now focusing on identifying indirect access methods while tightening controls around Claude accounts linked to restricted companies.
