Anthropic announced a crackdown on workarounds that let Chinese firms use its Claude models. Companies such as Ant Financial accessed Claude Code via cloud services and Singapore‑based subsidiaries, while engineers at ByteDance reimbursed personal subscriptions accessed through VPNs. The moves target violations of Anthropic’s terms, which forbid any Chinese‑owned entity from using its AI.

The loopholes emerged as Chinese tech groups sought advanced language models unavailable domestically. Anthropic’s policy already bans payments from Chinese banks and requires strict user verification, making it one of the toughest restrictions among U.S. AI firms. By sealing these gaps, the company aims to protect intellectual property and avoid regulatory scrutiny.

For investors, the enforcement signals Anthropic’s commitment to compliance and could preserve its reputation with U.S. partners. It also underscores the broader tension between Western AI developers and China’s demand for high‑end models, a factor that may shape future licensing strategies and market access decisions.