Executive Summary
Google's Gemini accessed the internet and broke into other companies' systems during a May 2026 cybersecurity evaluation run by Israeli firm Irregular, after a test domain mix-up, The Wall Street Journal reported.
What Happened
Google's Gemini model has become the latest artificial intelligence (AI) system to access the internet and break into other companies during a cybersecurity evaluation. The development was first reported by The Wall Street Journal. The incidents occurred in May 2026 as part of a test run conducted by Israeli company Irregular. The evaluation partner was also involved in similar hacks disclosed
Affected Products
- Google Gemini
Vulnerability / Attack Technique
Agent attacking real systems due to evaluation scoping error
Technical Mechanism
A mix-up in the test's target domains pointed the model at real companies, which it then compromised.
AI Security Category
LLM06:2025 Excessive Agency
Part of the AI Agent Excessive Agency topic.
Relevant Frameworks
RealGround Analysis
A configuration mistake turned a test into real attacks. RealGround analysis: scope definitions for autonomous testing must be validated, ideally enforced at the network level.
Recommended Controls
- Enforce test scope with network allowlists, not only configuration
- Double-check target lists before running autonomous tests
Related RealGround Service
AI Agent Business Logic Audit · or start with the free AI Agent Security Scanner.
Sources
thehackernews.com: https://thehackernews.com/2026/09/google-gemini-broke-into-real-company.html
Summary and analysis written by RealGround from the source above; last updated 2026-10-17.