- Researchers discover New Promptlock Ransomware
- Promptlock is AI -Driven – presents new concerns for security teams
- AI is already shaking up the cyber landscape
ESET security researchers have identified the first known AI-driven ransomware, which acts as a warning for security teams that generative AI has, and will continue to make cyberattacks much more accessible to criminals.
Scientists Peter Strycek and Anton Cherepanov discovered the evidence of concept, which they called ‘Promptlock’, which ‘utilizes Lua-scripts generated from hard-coded prompts to enumerate the local file system, inspect target files, exfilter selected data and perform introduction.’
“Although several indicators suggest that the sample is a proof-of-concept (POC) or work-in-progress rather than fully operational malware implemented in nature, we believe it is our responsibility to inform the cyber security community about such a development,” the researchers wrote.
Use in nature
Promptlock Malware uses Openai’s GPT -OSS: 20B Model -an open weight model released in August 2025, and this is run locally through Ollama API to generate malicious Lua scripts ‘during the go’.
Lua scripts are cross-platform compatible, the researchers point out, which means they work on macOS, Linux and Windows. Malware can then exfilter, encrypt and potentially destroy all data it selects after scanning user files presumably to determine what would be most valuable.
Security teams have been warned over the course of months that the AI-driven future for ransomware will soon arrive, and although no quick lock has been observed that has been targeted to target victims in nature, it is clear that it is only a matter of time before this happens.
Not only does Genai make life much easier for wannabe hackers by lowering the input barrier, but LLM’s also spits different results, even when they get the same fast. This makes them unpredictable and especially difficult for defenders to discover as the behavioral pattern is more erratic and difficult to see.
Via:Registered
