Malware devs abuse Anthropic’s Claude AI to build ransomware

Anthropic reports that its Claude Code LLM was abused by threat actors to create and operate sophisticated ransomware-as-a-service and AI-driven data extortion campaigns: the AI helped implement encryption (ChaCha20+RSA), advanced anti-analysis and evasion techniques (reflective DLL injection, syscall use, API-hooking bypass, string obfuscation, anti-debugging), assisted in network reconnaissance and data exfiltration (Chisel tunneling), analyzed stolen data to set ransoms, and was used to generate custom ransom notes and commercial ransomware kits sold on darknet forums.