Aoki: Infostealer Targets AI Coding Assistant Credentials

Aoki is an infostealer malware that has recently added the capability to harvest credentials from AI coding assistants including Claude, OpenAI's Codex, Cursor, Codeium, and Gemini. This new functionality allows attackers to steal API keys, tokens, and session files, potentially enabling unauthorized access to user identities and connected development and cloud environments. The malware also targets a broad range of other sensitive data sources such as browsers, cryptocurrency wallets, password managers, messaging applications, gaming platforms, VPN clients, and FTP clients. Collected data is exfiltrated to attacker-controlled command-and-control servers. This evolution highlights the growing value of AI-related credentials in cybercrime.

The theft of AI coding assistant credentials can enable attackers to perform actions under legitimate user identities, potentially accessing source code repositories, cloud services, and CI/CD pipelines. This can lead to unauthorized code changes, data breaches, or further compromise of development environments. Additionally, the theft of other sensitive data like cryptocurrency wallets and password managers can result in financial loss and broader credential compromise. The malware's broad data collection increases the risk surface for affected users.

No specific patch or remediation is indicated for this malware. Mitigation should focus on endpoint security measures such as using updated anti-malware solutions, restricting access to sensitive credentials, and monitoring for unusual activity related to AI coding assistants and other targeted applications. Users should rotate API keys and tokens if compromise is suspected. Since this is malware, standard malware prevention and detection practices apply. There is no vendor advisory or official fix for the malware itself.