On October 23, 2026, Check Point Research published “Infostealer riding the AI Train,” describing Aokigahara, a newly observed information stealer. The report notes that an earlier version may have been known as Kamidere or Stygian. Like other stealers, it targets browsers, VPN clients, cryptocurrency wallets, FTP tools, password managers, email apps and more. The samples discussed are Windows-oriented.
What stands out for AI tooling is the `aisessions` handler. Check Point says it collects authentication files, session data and configuration files associated with Claude (including Code／Desktop paths), ChatGPT Desktop, Codex, Cursor, Windsurf, GitHub Copilot, Gemini CLI and other developer AI tools. A separate `ide` handler gathers local editor and developer data—including Git credential files, SSH-related files under `.ssh`, and AWS／Azure／Google Cloud credential and config files—passing matching contents to the malware’s upload routine.
On monetization, the report stays careful: usable credentials or session tokens could enable unauthorized access to paid AI services or the resale of compromised accounts, and stolen access could also create data-extortion opportunities. At the same time, Check Point writes that while collection routines support these possibilities, the analysis does not confirm how operators monetize the stolen data. That is potential risk described in the research—not a finding that mass sales or confirmed extortion campaigns have been proven for this family.
This is a malware-capability report, not a notice that Claude, Codex, Cursor, Copilot or any other named product was “broken” by itself. We have not reproduced the samples; for reference only.