U.S. cybersecurity firm CrowdStrike says Chinese state‑linked actors accounted for more than half of all state‑sponsored intrusions against technology companies’ AI assets in the 12 months ending March 31. The focus has shifted from stealing hardware designs to exploiting human‑level vulnerabilities, such as insider access and social‑engineering campaigns amplified by AI‑generated content.
Anthropic has accused Alibaba of illicit attempts to pilfer its AI capabilities, while Copyleaks found that responses from Chinese startup DeepSeek’s R1 model closely resemble those of OpenAI’s ChatGPT, suggesting possible copying of U.S. models. A more personal case emerged when Agentiq Capital’s founder alleged a Chinese employee was a Beijing agent who altered code to sabotage venture‑capital funding, prompting an FBI complaint. Federal officials warn that such economic espionage costs the U.S. economy hundreds of billions of dollars annually.
Experts note that startups are especially exposed because they lack the security resources of larger firms, and that China’s policy push—free computing power and rent‑free office space—intensifies the race. The growing blend of technical and human‑focused attacks underscores the need for stronger defenses as the AI competition between the two superpowers accelerates.
