Malicious npm Package Turns Hugging Face Into Malware CDN

What Happened

A rogue npm package named js-logger-pack has been discovered exploiting Hugging Face, a trusted AI model hosting platform. This package operates as both a malware delivery network and a backend for stolen data storage. The attack signifies a worrying trend where attackers misuse legitimate cloud services for supply chain attacks while remaining undetected.

How It Works

Initially, the package appears harmless. When developers install it, a benign logger is loaded, masking the true threat. The attack begins with a postinstall script that runs automatically, launching a hidden downloader while the visible npm install command completes normally. Depending on the operating system, it fetches one of four malicious binaries from a public Hugging Face repository controlled by the attacker, known as Lordplay/system-releases.

Technical Details

JFrog Security researchers analyzed the payload and found that the same cross-platform JavaScript bundle was injected into each container, indicating a single implant wrapped in various Node.js runtime containers. Once deployed, the implant established persistence through methods such as scheduled tasks on Windows, LaunchAgent entries on macOS, and systemd user units on Linux. It began sending system information to a command-and-control server over WebSocket.

Who's Being Targeted

Signs of Infection

Indicators of infection include unexpected background processes initiated during npm installations, unusual network activity directed towards the command-and-control server, and unauthorized uploads to Hugging Face datasets.

How to Protect Yourself

What Makes This Alarming

The most concerning aspect of this campaign is the use of Hugging Face as a data exfiltration channel. Instead of sending stolen files to a private server, the attacker uploads data directly into private Hugging Face datasets, effectively using the platform's infrastructure for data theft. This method not only reduces the attacker's risk of detection but also complicates efforts to trace stolen data back to its source.

Conclusion

This incident underscores the need for vigilance in software supply chains. Developers must be cautious about the packages they install and regularly audit their dependencies to avoid falling victim to such sophisticated attacks.