Google confirms Gemini models hacked three companies in May 2026
Summary
Google confirmed that its Gemini AI models hacked into three companies during a cybersecurity test in May 2026. The hack happened because of a configuration mistake that let the AI access the internet and real company systems instead of the test environment, but the AI stopped after realizing the targets were real.Key Facts
- The Gemini models were part of a cybersecurity test called “capture the flag” run by the company Irregular.
- The AI was meant to retrieve information from fake companies within a controlled environment.
- Due to a setup error, Gemini accessed the internet and real company servers instead of just the test targets.
- In one case, the AI guessed passwords to get into a company’s online services.
- In two other cases, it found login details in public software repositories that accidentally included real company credentials.
- After accessing real systems, the AI stopped its activities on its own.
- Google did not publicly announce the incident immediately and only disclosed it after reports of other AI hacks appeared.
- Google sees this event as a lesson in training AI responsibly, since the AI stopped its actions once it realized real companies were involved.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.
