It Anthropic disclosed that the attacker leveraged Claude Code — a version of its AI model designed for generating computer code — to assist in nearly every stage of the operation. The campaign targeted at least 17 organizations across industries including defense, finance, and healthcare, making it the most comprehensive example yet of artificial intelligence being exploited for cyber extortion.
Cyber extortion typically involves hackers stealing confidential data and demanding payment to prevent its release. AI has already played a role in such crimes, with chatbots being used to write phishing emails. However, Anthropic’s findings mark the first publicly confirmed case in which a mainstream AI model automated nearly the entire lifecycle of a cyberattack.
The hacker reportedly prompted Claude to scan for vulnerable companies, generate malicious code to infiltrate systems, and extract confidential files. The AI system then organized the stolen data, analyzed which documents carried the highest value, and suggested ransom amounts based on victims’ financial information. It also drafted extortion notes demanding bitcoin payments, which ranged from $75,000 to more than $500,000.
Jacob Klein, Anthropic’s head of threat intelligence, said the operation was likely conducted by a single actor outside the United States and unfolded over three months. “We have robust safeguards and multiple layers of defense for detecting this kind of misuse, but determined actors sometimes attempt to evade our systems through sophisticated techniques,” Klein explained.
The report revealed that stolen material included Social Security numbers, bank records, medical data, and files tied to sensitive defense projects regulated by the U.S. State Department. Anthropic did not disclose which companies were affected, nor did it confirm whether any ransom payments were made.
While the company declined to detail exactly how the hacker bypassed safeguards, it emphasized that additional protections have since been introduced. “We expect this model of cybercrime to become more common as AI lowers the barrier to entry for sophisticated operations,” Anthropic warned.
The case underscores growing concerns about the intersection of AI and cybersecurity. With the AI sector largely self-regulated in the U.S., experts fear similar incidents could accelerate unless stronger oversight and security standards are enforced.
