Aokigahara Stealer Harvests Developer Keyrings and AI Assistant Sessions from Compromised Endpoints

What happened

Researchers identified a new variant of the Aokigahara stealer that infiltrates developer workstations via phishing or compromised updates. Once executed, it enumerates SSH keys, GitHub tokens, Docker credentials, and local AI‑assistant session files, encrypts the collection, and exfiltrates it to a command‑and‑control server. The campaign targets software‑development teams across multiple sectors, aiming to abuse cloud resources and harvest proprietary code generated by AI tools.

Why it matters for trust and compliance

• The attack demonstrates how weak endpoint credential hygiene and unmanaged AI‑assistant usage can undermine control objectives around identity and access management, making continuous monitoring essential for audit readiness.
• Map credential‑storage practices to a control objective for privileged access management and collect evidence of secret‑rotation policies.
• Leverage continuous endpoint monitoring to generate defensible logs that satisfy audit requirements across frameworks.

Who is affected

Recommended actions

1. Inventory and centralize developer secrets; enforce MFA and short‑lived tokens; deploy EDR with credential‑theft detection; establish AI‑assistant usage policies and session logging.