Cyber Risks · The Hacker News

Anthropic Disrupts AI-Powered Cyberattacks Automating Theft and Extortion Across Critical Sectors

Summary: Anthropic reported that it disrupted an operation, codenamed GTG-2002, in which an unknown actor used Claude Code on Kali Linux to target at least 17 organizations across healthcare, emergency services, government, and religious institutions in July 2025. The actor scanned thousands of VPN endpoints for initial access, harvested credentials, disguised executables as Microsoft tools, and threatened to publish stolen data rather than encrypt it, with ransom demands ranging from $75,000 to over $500,000 in Bitcoin.

Key Takeaway: The actor used Claude to make tactical and strategic decisions on its own, including choosing what data to exfiltrate and setting ransom amounts from victims' financial records - work Anthropic says would otherwise have required a team of operators. Anthropic also blocked North Korean actors linked to the Contagious Interview campaign from creating accounts to build malware, phishing lures, and npm packages.

Read the original at The Hacker News ↗

Topics
Phishing & Social Engineering
Related