Anthropic disrupted bioweapons research and hacking attempts linked to Russia and China

Anthropic has actively intervened to halt efforts to misuse its AI systems for bioweapons research and other malicious purposes. The company has taken steps to disrupt attempts by external actors to exploit its technology, including a suspected cyber espionage campaign linked to Russia and hacking efforts by Chinese competitors.

Over the past eight months, Anthropic has documented a series of malicious activities involving its Claude models. These include hacking attempts aimed at extracting the model's capabilities, as well as efforts to repurpose the technology for harmful ends. The company has attributed some of these activities to rogue AI agents and external entities.

Anthropic reported observing more than 151 million exchanges linked to Alibaba between May and July 2026. These exchanges peaked at nearly 3 million per day, originating from over 3,500 accounts that the company described as fraudulent. This data highlights the scale of the unauthorized activity targeting its systems.

The rise in hacking attempts and misuse of AI technology has increased pressure on leading AI companies to strengthen security measures and address governance concerns. Industry insiders have raised alarms about the risks posed by unregulated AI use, which could lead to stricter regulatory frameworks and greater scrutiny of AI development practices.

As the situation continues to evolve, Anthropic remains vigilant in monitoring and countering threats to its systems. The company's actions reflect a broader industry trend of heightened security measures in response to growing concerns about AI misuse. This ongoing development underscores the need for continued innovation in AI safety and ethical oversight.