The crazy, true story behind the first AI-powered ransomware

NYU researchers developed a proof-of-concept named Ransomware 3.0 showing how large language models can autonomously perform multiple ransomware lifecycle tasks—file discovery, targeted encryption, polymorphic payload generation, and personalized ransom notes. The lab-only binary was uploaded to VirusTotal and subsequently misreported as a real AI-powered ransomware (PromptLock) by third parties; the researchers and ESET clarified the sample is not operational outside a controlled environment, but the work highlights a realistic and concerning evolution of AI-assisted ransomware techniques.