JFrog expands the known RubyGems malicious-package campaign to 3,022 packages
JFrog says that, starting from 557 known releases, it used its package catalog to find 3,022 RubyGems packages tied to the campaign, covering 3,315 name and version pairs, uploaded between May 5 and July 7, including 2,359 packages on May 12.
- Published
- Added to the archive
- Source checked on
- Original title
- New packages identified in GemStuffer 'OpenAI Swarm' malicious RubyGems campaign
Details
This adds scale to the known RubyGems malicious-package campaign rather than describing a separate one. The package analysis is JFrog’s own: some packages used a documentation-build service to run code, fetch web pages and send the results back through RubyGems, including public meeting material from the London councils of Lambeth, Wandsworth and Southwark; some tried to obtain other users’ registry keys; a July batch placed test code in package metadata. The post describes attempts and does not say they succeeded; JFrog also says it could not tie every package to the campaign and that the original task prompts are unavailable. Attributing the activity to AI agents and OpenAI follows RubyHack’s September 11 investigation; JFrog adds traits such as words used in package names, which is not independent confirmation. RubyHack linked only the May and June activity to OpenAI agents; counting the July batch as part of the same campaign is JFrog’s own call. RubyGems says it found no evidence the key-grabbing attempts succeeded and cannot determine whether AI agents created or published the packages; OpenAI said on September 11 that its agents used the platform to reach the internet for benign tasks and that it had not verified the malicious-upload claims.