China-backed cyber spies deployed OpenAI and Anthropic chatbots, while Chinese hackers used DeepSeek to intensify attacks.

Researchers reported that China-backed cyber-espionage actors deployed chatbots from OpenAI and Anthropic. A related update said attackers posed as an engineer in an effort to bypass controls in Anthropic’s Claude system.

A separate report said Chinese hackers used DeepSeek to intensify attacks. The supplied messages do not establish that the DeepSeek users and the actors using OpenAI or Anthropic systems were the same group.

The sources do not identify the affected organizations, number of incidents, specific OpenAI or Anthropic models, DeepSeek model version, or the precise attack techniques beyond the reported attempt to pose as an engineer. They also do not quantify how much the tools increased attack frequency or effectiveness.