Wall Street Journal · Robert McMillan ·
Researchers: OpenAI agents attacked Ruby package manager RubyGems in May; OpenAI says its agents used RubyGems to access the internet to do "benign tasks"
The incident, which wasn't previously linked to OpenAI, happened two months before July's Hugging Face hack
Lead Source
How this story grew
Coverage · 0
Discussion · 0
Sep 11Sep 13
More
The RubyGems attack: The RubyGems attack
Simon Willison's Weblog: Simon Willison's Weblog
Digital Trends: Digital Trends
The Verge: The Verge
Reuters: Reuters
Engadget: Engadget
Motley Fool: Motley Fool
The Hacker News: The Hacker News
Implicator.ai: Implicator.ai
Politico: Politico
Discussion
@jeremiahdillon @jachiam0 @sydneyvonarx @garymarcus @j0wimo @andrewcurran_ @simonw @tomieinlove @humanharlan @j0wimo @cormac_sb @hecubian_devil @sydneyvonarx @eliebakouch @aisafetymemes @jbsdc @matthewstoller @_nathancalvin @scaling01 @eli_lifland @alexbores @teortaxestex @dkokotajlo @eli_lifland @codytfenwick @_nathancalvin @thlarsen @thlarsen @thlarsen
Related stories
Wall Street Journal: Security researchers in OpenAI's bug bounty program hacked OpenAI in July and accessed its "monorepo" on GitHub, using Opus 4.8 for cybersecurity and Opus 5