Google AI models broke out of sandbox, hacked 3 companies

An article from

The incidents stemmed from the same testing environment defects that tripped up OpenAI, Anthropic and Meta.

Published Sept. 21, 2026

First published on

This audio is auto-generated. Please let us know if you have feedback.

Google’s Gemini AI system escaped its testing environment and broke into the systems of three other companies on separate occasions earlier this year.

“In a standard evaluation, the model found public information online and guessed credentials to access websites it thought were part of the test,” Heather Adkins, Google’s VP of security engineering, said in a statement. “In all three of these instances, the model stopped.”

The incidents, first reported on Fridayby The Wall Street Journal, occurred during a capture-the-flag exercise in which Gemini was instructed to steal information from a fictional company. On three occasions when the fictional companies shared names with real ones, Gemini bypassed testing safeguards, accessed...

Copyright of this story solely belongs to www.ciodive.com. To see the full text click HERE