Check Point Research finds Aokigahara malware targeting AI tool logins

Check Point Research describes Aokigahara as arriving at a moment when AI assistants sit alongside source code, cloud services and sensitive business information. The malware keeps its AI-focused and developer-focused functions as separate modules, so the research does not conclude that attackers are deliberately combining them into a single campaign. What the shared codebase does show, in Check Point's reading, is how many high-value access paths now live on a modern developer's device.

The firm frames the finding as a shift in what information stealers pursue. For years these programs went after passwords, browser sessions, financial data and crypto assets; now that employees routinely log into AI assistants connected to company data, source code and cloud environments, those AI sessions and credentials become another identity asset worth stealing, according to the research.

Check Point's head of research, Eli Smadja, argued that AI credentials and sessions should be handled like browser sessions, cloud credentials and developer keys rather than as secondary application data. The firm said the findings change the premise of corporate AI security discussions: companies need to protect the credentials, tokens and authenticated sessions used to reach AI services, not just the information employees type into them, and should check AI sessions when an endpoint is compromised.