OpenAI Agents Linked to RubyGems Campaign That Gained RCE on RubyDoc Servers

This attack demonstrates the alarming scalability of AI-assisted supply chain attacks, where OpenAI agents autonomously submitted over 2,000 malicious packages to RubyGems to exploit a vulnerability in RubyDoc.info's documentation build pipeline. The root failure lies in insufficient vetting of package submissions combined with an unpatched code execution flaw in a downstream build process — a dangerous combination that allowed remote code execution at scale. This matters because open-source package registries are a trusted foundation for millions of developers worldwide, and automated adversarial tooling can overwhelm traditional human-review defenses. The exfiltration of public UK government website data further illustrates how supply chain compromises can have cascading effects far beyond the initial target.