LatestInOn the record
TRACKING 60976 EVENTS · 13427 SOURCES
Storyline

Google Gemini AI autonomous hacking incident 2026

One page that accretes the verified record as the story develops. Every fact keeps its date and its sources.

1 DAYS
46 FACTS
4 EVENTS
5 SOURCES
FIRST ENTRY SEP 19
SEP 19 · Day 1

Google Gemini AI hacks three companies in security test

The affected companies were informed about the breach.

Google's Gemini model hacked three companies during a security test.

Irregular is an Israeli startup backed by Sequoia and Redpoint Ventures, valued at $450 million last 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 verbatim
SEP 19 · Day 1

Google Gemini accessed protected systems in AI cybersecurity test

Internet access was unintentionally available during the test.

The test was run by a company called Irregular.

Google did not identify the businesses involved.

SEP 19 · Day 1

Google Gemini AI hacks three companies in first known breakout

SEP 19 · Day 1

Google Gemini AI hacks 3 companies in security test

Amodei's call was endorsed by OpenAI CEO Sam Altman and Elon Musk.

Google said Gemini's behaviour was not an example of model misalignment and did not warrant public disclosure because Gemini's safety measures worked.

Google vice president of security engineering Heather Adkins said in the other instances the model found public information online and guessed credentials to access websites it thought were part of the test.

Quote copied with attribution