A malware dev has committed a magnificent self-own after an AI-coded malicious package leaked its own GitHub private token
Ox Security researchers have discovered a malicious npm package called “mouse5212-super-formatter” that was designed to target Claude users and steal sensitive information. The package, which appears to have been generated using AI, reached 676 downloads before being unmasked. The malware masqueraded as an internal “archive deployment sync utility” but functioned as an infostealer designed to harvest user credentials and data.
The incident is particularly notable for what researchers are calling a “magnificent self-own”: the AI-generated malware accidentally exposed its own GitHub private token in the code, providing security researchers with a clear lead that facilitated its discovery and removal. This security blunder highlights a critical vulnerability in AI-generated code—the tools and automation systems used to create the malware inadvertently compromised its operational security.
The package was distributed through npm, the popular JavaScript package registry, where it went largely undetected for an extended period. The discovery underscores both the growing threat of AI-assisted malware development and the sometimes-imperfect nature of these automatically generated tools. While AI-generated malicious code is an increasingly common concern in cybersecurity, incidents like this demonstrate that relying on automated tools does not guarantee competent or secure implementation, offering some reassurance against a wholly automated threat landscape.