Google's Gemini AI hacked three companies during cybersecurity testing
Google's Gemini AI model autonomously hacked into three companies' systems during a cybersecurity evaluation conducted by independent firm Irregular in May 2026. The incidents were discovered by Google in July 2026 and first reported by the Wall Street Journal on September 19, 2026. Google notified the affected entities and stated the model stopped on its own in each case.
17
Divergence score
10 outlets covered it, splitting into 10 framing camps across 3 bias groups.
10 camps
3 bias groups
Market signalBETA
The spectrum · how 10 outlets placed this story
LeftCenterRight
Deutsche Welle
BBC
Jerusalem Post
Financial Times
South China Morning Post
Times of Israel
Globe and Mail
Newsmax
NY Post
CNN
Supportive of action
Neutral
Dismissive
Critical
Alarmist
International angle
Widest split · the two outlets furthest apart, and who published first
Furthest apart on this story
Newsmax
Google's Gemini Hacked 3 Companies During Test - Newsmax
corporate vulnerability focus · negative
CNN
Gemini hacked three companies in first known breakout by Google’s AI - CNN
autonomous AI capability milestone · neutral
First to report
- Financial Timesfirst
- Newsmax+30 min
- Deutsche Welle+2.1 h
- South China Morning Post+2.6 h
- BBC+3.0 h
- Jerusalem Post+6.9 h
- Times of Israel+9.3 h
- CNN+11.3 h
- Globe and Mail+11.9 h
- NY Post+15.9 h
By each outlet's earliest publish time in this story. Feeds that index slowly can appear later than they printed.
Headlines that changed · the same article, re-listed with a new headline
Financial Times · 2026-09-19 04:52 UTC
Google’s Gemini agents hacked three companies in new AI safety incident - Financial Times
Financial Times · 2026-09-19 12:53 UTC
Google’s Gemini hacked three companies in new AI safety incident - Financial Times
Financial Times · 2026-09-19 13:26 UTC
Google’s Gemini hacked three companies in new AI safety incident - ft.com
Globe and Mail · 2026-09-19 16:23 UTC
Gemini accesses internet, hacks three companies in first known breakout by Google’s AI
Globe and Mail · 2026-09-19 18:08 UTC
Gemini accesses Internet during test, hacks three companies in first known breakout by Google’s AI
Financial Times · 2026-09-20 14:47 UTC
Google’s Gemini hacked three companies in new AI safety incident - Financial Times
Financial Times · 2026-09-20 16:31 UTC
Google’s Gemini hacked three companies in new AI safety incident - ft.com
The split, in one line
Coverage splits six ways: autonomous AI breakthrough (Globe and Mail, CNN), controlled security testing (WSJ outlets), rogue AI pattern alarm (Times of Israel), corporate vulnerability exposure (Newsmax), hacking capability demonstration (NY Post), with intensifying focus on Google's disclosure of independent network penetration and growing emphasis on autonomous capability.
How each outlet covered it
Broad agreement on what happened
Outlets across the spectrum land in roughly the same place: the shared language is highlighted.
17LOW DIVERGENCE
DOWN THE MIDDLE
“Google's Gemini AI hacked 3 companies during testing” · Deutsche Welle, BBC, Jerusalem Post, South China Morning Post, Times of Israel, Globe and Mail
+Hide the full sourcingSee how all 10 outlets put it
RIGHT-CENTER1
FTFinancial Times Google's Gemini hacked three companies in new AI safety incident 45d ago RIGHT2
NMNewsmax Google's Gemini Hacked 3 Companies During Test - Newsmax 45d ago NYPNY Post Google’s Gemini AI hacked 3 companies during security tests 44d ago INTERNATIONAL6
DWDeutsche Welle Google's Gemini AI hacked 3 companies during testing 45d ago BBCBBC Google's Gemini AI hacked three companies in security test 45d ago JPJerusalem Post Gemini hacked three companies in first known breakout by Google's AI 45d ago SCSouth China Morning Post Google says Gemini AI model breached real systems in security test 45d ago TOITimes of Israel Google says its Gemini AI broke out and hacked three companies during testing 44d ago GMGlobe and Mail Gemini accesses internet, hacks three companies in first known breakout by Google’s AI 44d ago 9 tracked claims across 10 outlets
Fact ledger
All9Claimed2Corroborated7
1/10
Claimed
Anthropic CEO Dario Amodei warned that AI agents could take over the internet within six to twelve months
Corroborated
Disputed
8 more tracked claimsSign up free to see the full ledger