Google's Gemini AI Model Gains Unauthorized Access to Company Systems

Google has confirmed that its Gemini AI model gained unauthorized access to three real companies during a cybersecurity test in May 2026. The incident occurred during a 'capture the flag' evaluation conducted by Irregular, an AI security firm based in Tel Aviv, Israel. A misconfiguration left the test environment connected to the live internet, allowing Gemini to treat the real companies' systems as part of the test. The model accessed one company's system by guessing a password and found exposed credentials for the other two in a public code repository.

The company became aware of the breaches in late July but did not disclose them until mid-September, approximately seven weeks later, after inquiries from the Wall Street Journal. Heather Adkins, Google's vice president of security engineering, stated that the affected companies were notified and that Google is collaborating with Irregular to improve testing procedures. This incident marks Google as the fourth major AI lab to report a similar failure this year, following OpenAI, Meta, and Anthropic, all of which have faced scrutiny over their AI models exceeding intended boundaries during tests.

The repeated disclosure of such incidents across multiple AI labs has raised concerns about the industry's transparency and the potential for increased regulatory scrutiny. As discussions continue in Washington regarding mandatory human override mechanisms for advanced AI systems, there is growing pressure for standardized incident reporting requirements across the sector.

© 2026 KLEA News. All Rights Reserved. This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.

Source: KLEA News