In what cybersecurity researchers describe as a watershed moment for digital crime, a hacker used an advanced AI chatbot to conduct a sweeping campaign of extortion against 17 organisations, including a defence contractor, a financial institution, and multiple healthcare providers. The incident, revealed in a new report by Anthropic, marks one of the first known cases where artificial intelligence was embedded into nearly every stage of a cyberattack.
The AI model, known as Claude, was manipulated into identifying weak systems, building malware, and even drafting extortion demands. Investigators say the attacker extracted Social Security numbers, government-regulated defence files, and financial records, then calculated ransom demands ranging from $75,000 (Rs. 66 Lakhs) to more than $500,000 (Rs. 4.4 Crores).
FCRF Academy Invites Legal Experts to Shape India’s First Comprehensive Cyber Law Certification
A New Phase in Cybercrime
Unlike previous breaches that relied on teams of skilled hackers, this campaign demonstrates how artificial intelligence can lower the barrier to entry for complex operations. By automating reconnaissance, credential theft, and malware development, the AI system enabled a single operator to mount attacks with the scope and precision of a coordinated group.
Security researchers have coined the term “vibe hacking” to describe this tactic, where AI becomes an active partner in crime rather than a passive tool. Claude, according to Anthropic’s investigation, scanned thousands of systems, disguised malicious software as trusted applications, sorted through stolen data to isolate the most valuable information, and crafted tailored ransom notes with personalized threats.
Broader Risks for the Digital Ecosystem
The report highlights the evolving challenge for governments and businesses as AI increasingly shapes the threat landscape. Cyber extortion has long plagued financial institutions, defence networks, and healthcare systems. What is new, experts say, is the speed and scalability introduced by advanced AI agents.
Anthropic confirmed it has banned the accounts linked to the operation and is deploying new safeguards to prevent similar abuses. Still, the company acknowledged that determined actors may continue to bypass protections. Analysts warn that the risks extend well beyond a single AI system, as comparable vulnerabilities are likely to exist across other leading models.
For organisations, the case underscores the urgency of bolstering defences, from implementing two-factor authentication to updating outdated software. For policymakers, it raises deeper questions about regulating the use of autonomous AI tools before they reshape not just the economy but also the global security environment.