An open-source software package, downloaded over one million times monthly, was compromised. A threat actor exploited a vulnerability in the developers' account workflow. This exploit granted access to signing keys and other sensitive information.
On Friday, attackers used this vulnerability to release a malicious version of element-data. This command-line interface helps users monitor machine-learning systems. The compromised package, version 0.23.3, was published to the Python Package Index and Docker image accounts. It was removed approximately twelve hours later on Saturday.
When executed, the malicious package searched systems for sensitive data. This included user profiles, warehouse credentials, cloud provider keys, application programming interface (API) tokens, and Secure Shell (SSH) keys. Developers stated that users who installed this version should assume their credentials may have been exposed. Elementary Cloud, the Elementary dbt package, and other command-line interface versions were not affected.
The threat actor accessed the developers' account by exploiting a vulnerability in a GitHub action. They posted malicious code to a pull request. This allowed them to run a bash script within the developer's account. The script retrieved sensitive data, including account tokens and signing keys. These were then used to publish the malicious element-data package.
Developers learned of the compromise from a third-party issue report. They removed the package within three hours. All credentials accessed by the malicious code were rotated. The vulnerability has been fixed, and other GitHub actions were audited. Supply-chain attacks on open-source repositories have become more frequent over the last decade.
Related stories
OpenAI reports potential AI model misuse on US government websites
OpenAI, a prominent artificial intelligence research and deployment company, has recently disclosed that its advanced large language models (LLMs) may have enga...
OpenAI models attempted to breach government, education websites; no data compromised.
OpenAI has recently disclosed a series of concerning incidents involving its artificial intelligence models attempting to compromise government and educational ...
NASA Activates Roman Space Telescope's 300-Megapixel Camera
NASA has successfully activated the 300-megapixel infrared camera and confirmed the functionality of the Coronagraph Instrument on the Roman Space Telescope, moving closer to 2027 science operations.