MIT Tech Review: LLM-powered ransomware automates every attack stage with no human input

ChuckDBrooks · x · 2026-11-18

MIT Technology Review reports AI is already making cybercrime easier and could get much worse. In late August, ESET researcher Anton Cherepanov spotted a never-before-seen ransomware sample on VirusTotal that uses LLMs across every stage of an attack: calling an LLM to generate custom code in real time, rapidly mapping a victim's machine to find sensitive data to copy or encrypt, and writing personalized ransom notes — all autonomously. Researchers are split on whether AI-orchestrated attacks are already happening; some say it's too early to worry.

Original post →

More from Safety

Safety channel →