Attackers Target Developers With Fake DeepSeek Packages on PyPI
Threat actors are exploiting the recent hype around DeepSeek by distributing malicious Python packages designed to steal sensitive data. The packages are quickly removed, but developers still face significant risks from trendy, unverified code.
Attackers are exploiting the massive popularity of the DeepSeek AI model by publishing malicious packages to the Python Package Index (PyPI). Two packages, named deepseeek and deepseekai, pose as legitimate client libraries for the DeepSeek API but actually contain harmful code designed to steal sensitive information.
The malicious payload secretly collects user data, computer details, and environment variables that often hold critical credentials like cloud API keys and database passwords. Researchers note that the stolen data is sent to a command-and-control server hosted on Pipedream, and the malicious script itself shows signs of being written with the help of an AI assistant.
PyPI administrators quarantine the packages within 30 minutes of their initial publication, but 36 downloads still occur globally during that brief window. This incident highlights a growing trend of threat actors combining AI-generated malware with current tech hype, serving as a strong reminder for developers to carefully scrutinize newly released packages that claim to wrap popular services.