PromptLock uses a local AI model to generate Lua scripts compatible with Windows, Linux and macOS in real time. Based on predefined prompts, it autonomously decides whether to exfiltrate or encrypt data. It uses a model accessible via API and, despite being a proof of concept, represents a real threat
Researchers of ESET, a global European leader in the cybersecurity market, have discovered a new type of ransomware that uses generative artificial intelligence (GenAI) to carry out attacks. Called PromptLock, the malware uses a locally accessible AI language model to generate malicious scripts in real time. During the infection, the AI autonomously decides which files to search, copy or encrypt — marking a potential turning point in how cybercrime operates.
“The emergence of tools like PromptLock highlights a significant change in the cyber threat landscape,” said Anton Cherepanov, senior malware researcher at ESET, who analyzed the malware together with his colleague Peter Strýček.
PromptLock generates Lua scripts compatible with various operating systems, including Windows, Linux and macOS. The malware examines local files, analyzes their contents and — based on predefined text prompts — determines whether to exfiltrate or encrypt the data. A destructive function is already present in the code, although it is currently inactive.
The ransomware uses the 128-bit SPECK encryption algorithm and is written in Golang. The first variants have already been identified on the VirusTotal malware analysis platform. Although ESET considers PromptLock a 'proof of concept', the threat it poses is real.
“With the support of AI, launching sophisticated attacks has become much easier — you no longer need a team of expert developers,” Cherepanov added. "A properly configured AI model is now sufficient to create complex and adaptive malware. If implemented correctly, threats of this type could significantly complicate detection and make the work of cybersecurity defenders even more challenging."
PromptLock relies on a language model freely available via API, which means that malicious scripts are delivered directly to the compromised device. Interestingly, the prompt also contains a Bitcoin address, presumably linked to Bitcoin's creator, Satoshi Nakamoto.
ESET has published the technical details to raise awareness in the cybersecurity community. The malware has been classified as Filecoder.PromptLock.A.
