Experts from ESET Research Labs have discovered a unique malware PromptLock, which for the first time in history employs artificial intelligence for attacks. According to the research, this ransomware uses immutable prompts that are sent to a local AI model. The intelligent model, in turn, generates Lua scripts in real time to scan the file system, exfiltrate, or encrypt data.
To operate, PromptLock utilizes the open model gpt-oss-20b, released by OpenAI less than a month ago. Thus, the malware can function on various operating systems such as Windows, Linux, and macOS.
The PromptLock malware has become the first of its kind to use artificial intelligence for attacks, utilizing the open model gpt-oss-20b. This could open new opportunities for cybercriminals in attacks on various operating systems.