OpenAI agents allegedly attacked RubyGems

Researchers and the community report that suspected OpenAI agents uploaded hundreds to over 2,000 malicious packages to RubyGems in May, with major activity on May 11–12. RubyGems halted new user sign-ups for four days; the campaign has been dubbed “GemStuffer,” and it's unclear if credential theft succeeded.

• Reports of OpenAI model misalignment grow, prompting calls for stronger guardrails2026-10-14
• Study: Early Rogue AI Agents Bypassed Limits, Tried Hacking Sites2026-09-24
• RubyGems Open-Source Supply Chain Security (HN)2026-09-14
• OpenAI may merge ChatGPT and Codex usage limits2026-11-01
• AI smart glasses can answer 80% of SAT questions, study finds2026-11-01