Researchers from ESET have identified PromptLock, the first known AI-powered ransomware, which is currently a non-functional proof-of-concept. This prototype utilizes OpenAI's gpt-oss-20b model to generate malicious Lua scripts and operates within a controlled environment, highlighting the potential dangers of AI in cybercrime despite no active infections being reported.

NYU researchers developed a proof-of-concept AI-powered ransomware, dubbed Ransomware 3.0, which utilizes large language models to create customized attacks targeting specific files on victim systems. The project unexpectedly gained attention when security analysts mistakenly identified it as a real threat, prompting discussions about the implications of AI in ransomware development. While the malware is not functional outside a lab setting, researchers warn that the techniques could inspire actual cybercriminals to create similar threats.