Anthropic warns of AI-driven hacking activity linked to China

A team of researchers has revealed that this is the first reported use of artificial intelligence to direct a hacking campaign in a near-automated manner.

AI company Anthropic announced this week that its researchers had disrupted a cyber operation it said was linked to the Chinese government. The operation involves the use of artificial intelligence systems to direct the hacking campaign, which researchers call a disturbing development that could significantly expand the attack surface of AI-powered hackers.

Concerns about the use of AI to facilitate cyber operations are not new, but a concern with new operations is the extent to which AI has been able to automate some tasks, the researchers said.

“While we expected these capabilities to continue to evolve, what caught our attention was how quickly they were happening at scale,” they wrote in the report.

The operation was small in scope, targeting only about 30 people working at technology companies, financial institutions, chemical companies, and government agencies. Anthropic became aware of the operation in September and took steps to halt the operation and notify affected parties.

Anthropic says hackers have been “successful in only a minority of cases” and notes that while AI systems are increasingly being used in a variety of work and leisure settings, they could also be weaponized by hacking groups working on behalf of foreign adversaries. Anthropic, maker of the generative AI chatbot Claude, is one of many tech companies touting AI “agents” that access computer tools and perform actions on behalf of people, beyond the capabilities of chatbots.

“Agents are valuable to day-to-day operations and productivity, but when exploited they can significantly increase the likelihood of large-scale cyber attacks,” the researchers concluded. “These attacks could become even more effective.”

A spokesperson for the Chinese embassy in Washington did not immediately return a message seeking comment on the report.

Microsoft warned earlier this year that foreign adversaries are increasingly employing AI to make cyberattacks more efficient and less labor-intensive.

America’s adversaries, as well as criminal organizations and hacking companies, have harnessed the potential of AI to automate and improve cyberattacks, spread inflammatory disinformation, and infiltrate sensitive systems. AI can, for example, translate poorly worded phishing emails into fluent English or generate digital clones of government officials.

© 2025 Associated Press. Unauthorized reproduction is prohibited. This material may not be published, broadcast, rewritten or redistributed without permission.