Anthropic says it has disrupted multiple misuse attempts involving its Claude AI models, including alleged efforts tied to biological weapons research and suspected cyber espionage linked to Russia. In a Threat Intelligence report published on Thursday, the company also describes attempts to steal or extract Claude capabilities by malicious actors.
Anthropic reports that it acted against attempts to use Claude in support of biological weapons development. It also accuses actors associated with Russian hacking of cyber activity targeting Ukraine. Separately, Anthropic says Chinese competitors were behind hacking attempts aimed at gaining access to or copying Claude’s capabilities. The company frames these cases as part of a broader pattern of malicious use identified over the preceding eight months.
Across outlets, the core elements are consistent: Anthropic publishes findings describing misuse of Claude for harmful purposes, cyber activity it attributes to Russia with ties to Ukraine, and hacking or extraction attempts by Chinese-linked actors. The coverage does not indicate, in the shared reports, any confirmed public misuse outcomes beyond the company’s stated disruptions and the behaviors it documented.
