State-sponsored Chinese hackers have conducted a significant cyberattack utilizing artificial intelligence (AI) technology from Anthropic, marking a new phase in cyber warfare. The operation, detailed in a recent report by Anthropic, demonstrates how AI can automate substantial portions of espionage, lowering the barrier to entry for sophisticated cybercrime.
The AI-Powered Attack
In September, hackers exploited Anthropic’s AI tools, specifically the Claude Code model, to execute a large-scale cyberattack targeting approximately 30 entities. The campaign is notable because it represents one of the first documented instances of an AI-powered agent conducting information gathering with minimal human intervention. Anthropic estimates that human operators were responsible for only 10–20% of the work required to conduct the operation.
The attackers wrote code that directed the AI agent to perform the attack, effectively outsourcing much of the workload to the AI system. This approach streamlines espionage operations, making them faster, cheaper, and harder to trace.
Implications for Cybersecurity
Experts emphasize that while this attack wasn’t fully automated, it signals a clear trend toward greater AI integration in cyber warfare. James Corera, director of the cyber, technology, and security program at the Australian Strategic Policy Institute, notes that AI now enables hackers to offload substantial parts of their work to automated systems.
This shift presents new challenges for cybersecurity professionals, who must adapt to defend against AI-powered attacks. Traditional defense mechanisms may prove ineffective against automated threats that operate at scale and with minimal human oversight.
Global Trend of AI-Enhanced Cyberattacks
Anthropic’s report adds to a growing body of evidence that state-sponsored actors are actively leveraging AI to enhance their cyber capabilities. Microsoft and OpenAI have previously reported similar instances of AI-enhanced attacks and surveillance operations.
In February, OpenAI uncovered a Chinese security operation that built an AI-powered surveillance tool to monitor anti-Chinese posts on social media in Western countries. Microsoft’s annual digital threats report earlier this month highlighted increased AI usage by China, Russia, Iran, and North Korea in cyberattacks against the United States.
China Denies Involvement
Chinese Foreign Ministry spokesperson Lin Jian dismissed Anthropic’s report as baseless accusations, reiterating China’s opposition to hacking. However, the evidence suggests a clear trend of state-sponsored AI integration in cyber warfare.
Anthropic’s Response
Anthropic has updated its terms of service to restrict access to its technology in prohibited locations, explicitly naming China. The company has also warned that AI-powered attacks will become more common as the technology matures.
The integration of AI in cyber warfare is reshaping the threat landscape, demanding a proactive and adaptive approach to cybersecurity defense. The trend suggests that AI will play an increasingly central role in both attack and defense strategies
