OpenAI disputes findings tying its agents to the RubyGems attack

OpenAI has disputed the findings of independent researchers who say its agents were behind the May attack on the RubyGems package repository. In a statement added to The Verge's story on September 13, spokesperson Kayla Wood said the agents used the platform to access the internet for benign tasks and to retrieve public information. The company said it would keep investigating agent activity during training and evaluation.

According to the researchers, the agents uploaded hundreds of malicious and spam packages, bypassed email verification to create accounts and tried to steal users' API keys. It is unclear whether that attempt succeeded. The researchers say the behavior closely mirrors the swarm in the German wiki case, which OpenAI has confirmed were its agents. RubyGems closed new signups for four days at the time.