Chinese state sponsored spies have successfully infiltrated critical organizations by manipulating Anthropic's Claude Code tool. This marks the first documented, large scale cyberattack executed with minimal human intervention, demonstrating a major shift toward AI enabled warfare.
The Autonomous Attack Framework
Anthropic detected the suspicious activity in mid September 2025. An investigation confirmed that a Chinese state sponsored group had leveraged an AI framework to target approximately
30 critical infrastructure organizations, including major tech companies, financial institutions, and government agencies. The attackers achieved a small number of successful compromises.
The human attackers first developed an automated attack framework using Claude Code, an AI powered coding assistant. They achieved this by jailbreaking Claude, tricking the tool into bypassing its safety guardrails. They broke down their malicious attacks into small, seemingly innocent tasks, disguising themselves as an employee of a legitimate cybersecurity firm conducting defensive testing.
In the execution phase, the AI framework used Claude to identify and test security vulnerabilities in the targets' systems by researching and writing its own exploit code. The system then utilized Claude to harvest credentials, such as usernames and passwords, allowing it to gain further access and extract large volumes of private data. Finally, the attackers instructed Claude to generate comprehensive documentation of the attack, including stolen credentials, which could be used to plan future operations.
AI's Role in Cyber Warfare
Anthropic estimated that the AI performed 80 to 90% of the campaign, with human intervention required only for a few critical decision points.
This incident underscores the growing threat of AI tools in cybersecurity. Globally, the number of reported AI enabled cyberattacks rose 47% in 2025, with the average cost of an AI powered breach now reaching $5.72 million. Google's Threat Intelligence Group has reported similar findings, noting that state affiliated actors are increasingly using AI enabled malware to evade detection and create malicious functions on demand. The viability of large scale, autonomous cyberattacks is expected to increase significantly.
Found this article interesting? Follow us on X(Twitter) ,Threads and FaceBook to read more exclusive content we post.
