PromptLock: First AI-Powered Ransomware Emerges

PromptLock is the first known AI-powered ransomware to rely on AI systems for local operations, though it is described as a proof-of-concept or work-in-progress and not deployed in the wild yet, according to ESET. Written in GoLang, the malware relies on OpenAI’s GPT-OSS:20b, an open-weight model that can be used without proprietary restrictions, to generate Lua scripts on the fly that perform tasks such as filesystem enumeration, file inspection, data exfiltration, and encryption.

Both Windows and Linux variants have been observed, with the generated Lua scripts being cross-platform compatible and the ransomware using the SPECK 128-bit algorithm for file encryption. The concept requires several pre-conditions, including running Ollama locally to access the GPT-OSS:20b model via the Ollama API, and it may exfiltrate, encrypt, or potentially destroy user data based on detected files, though destruction appears not yet implemented.

ESET notes that the malware’s success would also hinge on poor network segmentation, lax prompt guardrails, or outgoing traffic to large language model ports and protocols.