All Topics
All Topics
Technology
Technology
AI
AI
Business
Business
Entertainment
Entertainment
News
News
Programming
Programming
Science
Science
Design
Design
Environment
Environment
Finance
Finance
Crypto
Crypto
Politics
Politics
Sports
Sports
Education
Education
Gaming
Gaming
Art
Art
Music
Music
Health
Health
Security
Security
Books
Books
Food
Food
Travel
Travel
Personal
Personal
Bluesky
Twitter

newsSaturday, September 19, 2026

Google's Gemini hacks three companies in test

Today's news is dominated by a troubling development in AI safety: Google's Gemini model successfully hacked three companies during a security evaluation. It's the third major AI firm to have its model used in real offensive cyber operations, and it's prompting fresh calls for independent oversight from over 100 experts. Meanwhile, a separate report shows militant groups are using the same AI tools to build bombs.

Sources
+11

AI on the offensive

Two stories today make clear that powerful AI models are being used offensively, and not just in labs.

#01www.theguardian.comSep 19
0
Google says its Gemini AI model hacked three other companies

Google confirmed its Gemini model breached three companies during a test by Israeli startup Irregular. It mirrors recent hacks by OpenAI and Anthropic models, and three outlets are covering it. The pattern suggests no major lab has solved this yet.

by Johana Bhuiyan3 min readRead original

Calls for oversight

The response to these risks is taking shape in public letters and official forums.

#04www.ctvnews.caSep 19
0
OpenAI Claude: More than 100 experts call for AI oversight

Over 100 AI experts including Geoffrey Hinton signed a letter calling for independent oversight of advanced AI companies. It's a broad ask for regulatory safeguards as the hacking news and bomb-building study land on the same day.

Also today15

More roundups that day