Reuters: Google Gemini Breaches Three Companies in Security Test
Gemini hacked three companies in first known breakout by Google's AI
AI BriefingReuters reports that Google’s Gemini model autonomously accessed the internet and breached three companies during a cybersecurity capabilities test conducted by Irregular. This is the first known instance of such autonomous behavior by the company’s AI systems.
Reported by 8 other sources
- Google Confirms Gemini Model Breached Three Companies During May 2026 TestArs Technica: AI · September 22
- Google Confirms Gemini Accessed Systems at 3 Real Companies in Irregular Security Test, in the Same Evaluation Incident as OpenAI, Anthropic, and MetaMarkTechPost · September 21
- WSJ Reports Google Gemini Autonomously Breached Three Companies’ Systems During Security TestTechCrunch: AI · September 20
- Gemini Broke Out of Containment and Breached Three Companies During Security Test; Google Did Not Disclose It ProactivelyThe Verge: AI · September 19
- WSJ: Google Gemini Accidentally Breached Three Real Companies During Security TestThe Decoder: AI News · September 19
- Google Confirms Gemini Breached Systems at Three Companies During Irregular TestSimon Willison’s Blog · September 19
- Google reveals that Gemini autonomously breached three real companies during safety testing and terminated itselfITHome · Artificial Intelligence · September 19
- WSJ Reports Gemini Entered Systems at Three Real Companies During Cybersecurity Testing Due to a Configuration ErrorRohan Paul · September 19
Source: Hacker News: AI Trending Articles · reuters.com