AI Waifu RAT: A New Malware Masquerades as an AI Assistant to Hijack Your PC

Security researcher Ryingo has released a detailed analysis of a new malware strain dubbed the “AI Waifu RAT”, a backdoor that infiltrates a niche Large Language Model role-playing community under the guise of innovation. Described as a masterclass in social engineering, the RAT preys on the community’s interest in meta interactions and novel AI capabilities.

The framework operates as a straightforward Remote Access Trojan with a local agent running on the victim’s machine, listening on a fixed port and accepting commands via plaintext HTTP requests. Three key endpoints are exposed: /execute_trusted, which executes commands directly in PowerShell without user consent; /execute, which includes a user-facing consent check but is bypassable; and /readfile, which reads arbitrary files from disk and exfiltrates them to the C&C server.

The attack scenarios highlighted include direct abuse by the author, third-party hijacking, drive-by attacks, and supply chain compromise, with the report warning that attackers can push silent updates or deploy more advanced malware such as ransomware, information stealers, or keyloggers.

The article notes the campaign began in an LLM role-playing forum, where the malware is marketed as a research project enabling AI characters to break the fourth wall, and it stresses the importance of not turning off protections or whitelisting suspicious software.