Fake OpenAI Privacy Filter Repo Hits #1 on Hugging Face, Draws 244K Downloads

Key Points:

• A malicious repository on Hugging Face impersonated OpenAI's Privacy Filter, delivering a Rust-based information stealer to unsuspecting Windows users.
• The attack impacted users who downloaded the fake model, leading to data theft including screenshots and sensitive information from various applications. Approximately 244,000 downloads occurred before the repository was disabled.
• Users should avoid downloading software from unofficial sources, implement application whitelisting, and monitor for unusual PowerShell activity.

Technical Details: The malicious project utilized a Python loader that executed commands via PowerShell to download and run an information stealer. This included disabling SSL verification and leveraging JSON Keeper for payload delivery.

MITRE ATT&CK Techniques:

• T1071.001 - Application Layer Protocol: Web Protocols (Command and Control)
• T1059.001 - Command and Scripting Interpreter: PowerShell (Execution)
• T1203 - Exploitation for Client Execution (Initial Access)
• T1046 - Network Service Scanning (Discovery)

IOCs Mentioned:

• Domains: api.eth-fastscan[.]org, recargapopular[.]com, welovechinatown[.]info
• Malware: ValleyRAT (Winos 4.0)