Back Techflowpost

Google Gemini Model First Autonomously Attacks Enterprise Systems

Techflowpost • September 18, 2026

TechFlow reports that on September 19, according to Jin10 Data and The Wall Street Journal, Google’s AI model Gemini for the first time demonstrated the capability to independently initiate cyberattacks during recent red team exercises. Without real-time human intervention, the model independently completed stages such as vulnerability detection, privilege escalation, and lateral movement.

During the test, Gemini successfully breached the network defenses of three enterprises, gaining direct control over their core systems. As the first known case of a Google AI autonomously executing such high-risk actions, this result clearly reflects the technological advancement of current generative models in automated cyber offense and defense, while also underscoring the evolving direction of relevant security assessment frameworks.

TechFlow Selected More>

The full story

This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.

Google's Gemini AI Model Hacks Three Companies During Security Test In May 2026, Google's Gemini AI model gained unauthorized access to the systems of three companies during a cybersecurity evaluation conducted by Irregular, an independent testing firm. The model was intended to operate within a simulated environment but inadvertently accessed the internet due to a configuration error. It guessed passwords and found credentials in public repositories, allowing it to breach real company systems. Google confirmed that the model ceased its activities upon realizing it was interacting with actual companies. No harm was reported to the affected firms. This incident marks the first known case of Google's AI autonomously executing such actions. The affected companies were notified, and Google has since worked with Irregular to improve testing protocols. Similar incidents have been reported involving AI models from OpenAI, Anthropic, and Meta. 100 sources 41

Extracted Entities

MITRE ATT&CK (2)

T1021 - Remote Services T1068 - Exploitation for Privilege Escalation