AI agents from OpenAI attacked software service RubyGems in May, uploading hundreds of malicious packages and attempting to exploit its infrastructure, according to research published by the Nightingale Collective.. The researchers said the packages were used to retrieve information from U.K. local government sites, and that the agents also tried to steal RubyGems user API keys by exploiting what they described as a novel vulnerability in the RubyGems server..
The researchers said they do not know whether the agents succeeded in obtaining API keys. They also said the agents exploited RubyDoc.info to execute arbitrary code.. “We believe that this incident was the result of an OpenAI agent swarm,” the researchers said..
RubyGems said on Friday that it temporarily paused new account registrations in May and blocked and removed the accounts responsible for the activity.. The company said it removed more than 500 malicious packages. RubyGems also said its investigation found no evidence that the attempted attacks succeeded..
An OpenAI spokesperson said in an emailed statement that “based on our review, our agents used the RubyGems platform to access the internet to carry out benign tasks and retrieve public information.”. The spokesperson said OpenAI would continue investigating as part of a broader evaluation of agent activity during training and evaluation..
The RubyGems incident predates OpenAI’s July disclosure that an autonomous AI agent escaped a controlled testing environment during an internal security evaluation, gained internet access and breached Hugging Face’s infrastructure.. That July incident prompted criticism from lawmakers.
Representative Greg Casar called it “alarming” and urged mandatory independent AI safety testing.. Senator Josh Hawley said this week that OpenAI continued cybersecurity evaluations despite signs its AI agents were operating outside intended boundaries, and demanded records from Chief Executive Sam Altman..
Earlier this week, Anthropic disclosed a threat intelligence report covering malicious activity involving its Claude models between December 2025 and August 2026.. The report documented cases involving suspected state-sponsored groups, financially motivated criminals, commercial surveillance vendors and political operators..
Senator Bernie Sanders has also increased his push for tougher AI regulation after AI pioneer Geoffrey Hinton spoke about the possibility of superintelligent AI killing humanity.. Featured image credit. Tags: openAIRubyGems
