A Chinese think‑tank envoy met with Anthropic in Singapore last month, urging the U.S. startup to grant Beijing access to its Mythos model. Anthropic refused, citing security concerns. The exchange, though informal, signals Beijing’s push to close the AI gap.

Anthropic’s Mythos, unveiled in April, can uncover software vulnerabilities and has been withheld from public release. The model was shared only with the U.S. government and 40+ firms, a move that underscores the company’s focus on national‑security clients.

U.S. officials see the incident as part of a broader AI rivalry that mirrors Cold War tensions. Analysts estimate that the latest U.S. models push China’s lead back by nine months to a year, widening the technology gap.

The meeting highlights how AI developments are now treated as strategic assets. China’s refusal to accept foreign investment without approval and its blocking of Meta’s $2 billion purchase of Manus illustrate a tightening of controls. Anthropic’s stance may dampen cross‑border collaboration and could influence future AI‑related trade policies.