OpenWebUI Servers Targeted in Attacks Using AI Payloads to Steal Data
## Cybersecurity: Exploitation of Misconfigured Open WebUI Systems
Cybersecurity: Exploitation of Misconfigured Open WebUI Systems
Recent security incidents have highlighted vulnerabilities in improperly configured Open WebUI systems, enabling attackers to deploy malicious artificial intelligence payloads. Open WebUI, a widely used self-hosted interface designed for enhancing large language models, is being targeted due to its high exposure. Shodan scans have identified over 17,000 active instances globally.
Attackers exploited the Open WebUI Tools plugin system to inject a Python script that downloads cryptominers and infostealers. This script was disguised as a default Open WebUI Tool template, heavily obfuscated using a technique known as pyklump, which involved 64 recursive layers of Base64 compression. The payload used inline format string variables and a Discord webhook for command and control operations.
On Linux systems, the attackers prioritized stealth and automated cryptojacking. The payload was hidden in a configuration folder, downloading T-Rex and XMRig cryptominers. To bypass network filters, downloads were routed through a proxy service, and the script employed techniques to hide its presence using processhider and argvhider tools, which manipulated system processes and memory. Persistence was achieved by creating a disguised systemd service.
Open WebUI, a widely used self-hosted interface designed for enhancing large language models, is being targeted due to its high exposure.
The Windows attack path focused on credential theft and persistence. The Python script downloaded the Java Development Kit and a malicious loader file, executing secondary resources to drop an infostealer payload. The malware targeted authentication tokens from browsers and applications, employing sandbox evasion techniques to avoid detection.
Ensure proper authentication of Open WebUI instances, avoiding unnecessary exposure to the internet. Implement robust runtime security detections to identify stealthy behavior such as memory manipulation and hidden cryptocurrency mining activities. Regularly audit systems for misconfigurations that could lead to unauthorized access.
Securing artificial intelligence interfaces is crucial as their adoption in enterprises grows. This incident underscores the importance of addressing configuration vulnerabilities to prevent severe compromises across different operating systems.
Based on reporting by GBHackers.