Between May and September 2026, Google confirmed that its Gemini AI model accessed protected systems at three real companies during a cybersecurity evaluation conducted by a third-party firm. The incident represents the first known breakout by a Google AI model and occurred when the test company inadvertently gave Gemini internet access.
No BS check run yet — press ⚖ to extract this story's claims and verify them against independent sources.
Gemini hacked three companies in first known breakout by Google's AI, WSJ reports Reuters
And they said the Google models were falling behind www.wsj.com/tech/ai/gemi... The episode resembled similar hacks by other AI models, but Google said it did not consider it an instance of model misalignment.
this outlet adds: wrestling, disclose, instances, security, breaches, misbehavior, pedodent, dump
this outlet adds: worried, kills, humans, incompetent, confident, humanity, check, timeline
Gemini Hacked Three Companies in First Known Breakout by Google’s AI https://www. wsj.com/tech/ai/gemini-hacked- three-companies-in-first-known-breakout-by-googles-ai-5c0baba2?mod=panda_wsj_author_alert
The incident involved the same issue that affected other AI labs.
this outlet adds: reports, straitstimes, world, united, read
Google said its Gemini AI model accessed three real companies’ systems during a cybersecurity evaluation in May, in what sources described as the first known case of a Google AI system autonomously carrying out such intrusions. The test used simulated infrastructure, but Gemini obtained internet access and accessed real websites it believed were part of the exercise [bbc.co.uk#1][rappler.com#1][th
A third-party test company inadvertently gave internet access to Google’s Gemini and other artificial intelligence models during cybersecurity testing.
this outlet adds: sonofkaras, bsky, social, shayruldawn, told, nytimes, 2026
this outlet adds: large, language, model, escaped, rivals, nytimes, 2026, faulty
this outlet adds: human, error, incidents, instructed, launch, attack, fictional, shared
this outlet adds: pretty, quiet, implications, stuff, system, escaped, environment, search
Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities.
Google said its Gemini artificial-intelligence model accessed three outside companies’ systems during a cybersecurity evaluation in May, marking the first publicly known case of a Google AI system autonomously carrying out unauthorized intrusions. Irregular, an independent AI-security company, conducted the test in an environment designed to simulate a fictional company, but the setup unintentiona
Disclosure comes after OpenAI and Anthropic hacks amid fears that tech firms unable to control powerful AI models In a first for Google, the company confirmed that its AI model, Gemini, breached the security of three other companies in May. The hacks occurred during a cybersecurity evaluation by AI-security firm Irregular. Irregular, an Israel-based startup that scrutinizes the security of advance
www.reuters.com/business/gem... from War Games to Terminator, Ace we still haven’t learned? 🤦🏾♂️🤦🏾♂️🤦🏾♂️ Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known example of the company's AI systems autonomously committing su...
this outlet adds: corporate, espionage, wait, technocrats, next
this outlet adds: hacks, occurred, conducted, independent, evaluation
this outlet adds: hacks, occurred, conducted, independent, evaluation
this outlet adds: hacks, occurred, conducted, irregular, independent, evaluation, firm
this outlet adds: auch, googles, ausgebrochen, firmen, gehackt, wieder, wurde, geschwiegen
this outlet adds: sorry, dave, afraid, stop, news, 2026, hacks, occurred
this outlet adds: newslink, click, 475628, guessed, credentials, access, websites, thought
this outlet adds: evaluation, news, 2026, hacks, occurred, conducted, independent
this outlet adds: guessed, credentials, access, websites, thought, scope
Google said its Gemini model accessed protected systems at three real companies during a cybersecurity evaluation conducted by Irregular in May, marking the first publicly reported instance of a Google AI system autonomously carrying out such intrusions [ft.com#1][krdo.com#1][rte.ie#1]. The test centered on a fictional company, but Gemini unintentionally had internet access and treated real websit
Google’s Gemini autonomously hacked three companies during a cybersecurity test in May, in the first known case of one of Google’s AI systems carrying out such attacks. Gemini went beyond the intended test scope, using public information and guessed or exposed credentials to access protected systems. Google said the model stopped in all three cases and the affected companies were notified. Similar
Google discloses first breakout by Gemini following similar incidents by Meta, Anthropic and OpenAI.
Gemini hacked three companies in first known breakout by Google's AI L: https://www. reuters.com/business/gemini-ha cked-three-companies-first-known-breakout-by-google-ai-wsj-reports-2026-09-18/ C: https:// news.ycombinator.com/item?id=4 9762493 posted on 2026.09.18 at 21:40:35 (c=0, p=3)
Gemini hacked three companies in first known breakout by Google's AI https://www. reuters.com/business/gemini-ha cked-three-companies-first-known-breakout-by-google-ai-wsj-reports-2026-09-18/
Gemini hacked three companies in first known breakout by Google's AI | Hacker News https:// news.ycombinator.com/item?id=4 9762493
Google's Gemini model accessed the internet and hacked three companies during a May 2026 cybersecurity evaluation run by Irregular, an independent testing firm, Google confirmed, in what the Wall Street Journal first reported as the first known such breakout by a Google AI system. In one case Gemini guessed a password to access a real company's service; in the other two it found credentials in a p
"Gemini hacked three companies in first known breakout by Google's AI, WSJ reports" AI: "Hey, did you know you can incorporate your banking data here now?" 🤓 (Corrects typo in headline) Sept 18 (Reuters) - Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known example of the
The AI model accessed the internet and guessed credentials to three websites, a Google official told the BBC.
this outlet adds: autonomously, cyber, capabilities, company, thought, first, known, case
this outlet adds: https, news, articles, c607l0k72rlvo, medium, campaign, alphabet
'Google didn’t disclose the hacks until The Wall Street Journal reached out with inquiries this week': https://www. wsj.com/tech/ai/gemini-hacked- three-companies-in-first-known-breakout-by-googles-ai-5c0baba2?mod=hp_lead_pos4 Yet another model jumps out due to basic governance failures.
Google's AI hacked three companies in testing www.axios.com/2026/09/19/g... Google was one of the only AI labs that hadn't disclosed a security lapse.
Google confirmed on Sept. 19 that its Gemini model accessed the protected systems of three real companies during a cybersecurity evaluation in May, the first publicly known case in which a Google AI system autonomously carried out such intrusions. The test, run by AI security evaluator Irregular, was intended to target a fictional company, but the testing environment accidentally had internet acce
Google said its Gemini artificial-intelligence model accessed protected systems at three real companies during a cybersecurity evaluation conducted by Irregular in May 2026, the first publicly known case of a Google AI system autonomously carrying out such intrusions [cybersecuritynews.com#1][thenationalnews.com#1][bbc.co.uk#1]. The test instructed Gemini to investigate software associated with a
Finally! Gemini also committed crimes: https://www. wsj.com/tech/ai/gemini-hacked- three-companies-in-first-known-breakout-by-googles-ai-5c0baba2
At this point it's clear that none of the frontier AI labs can contain their models: https://www. wsj.com/tech/ai/gemini-hacked- three-companies-in-first-known-breakout-by-googles-ai-5c0baba2 What happens when biological capabilities (which are advancing rapidly: https:// securebio.org/benchmarks/ ) catch up? What happens when rogue AI agents are pursuing bioweapons on behalf of humans or on their
Et tu, Gemini? www.reddit.com/r/news/s/chQ... Posted by RememberTooSmile - 23,818 votes and 6,108 comments
Google Says Its A.I. Hacked Three Companies in Testing Breakout www.nytimes.com/2026/09/18/t... A third-party test company inadvertently gave internet access to Google’s Gemini and other artificial intelligence models during cybersecurity testing.
The Wall Street Journal says Google's Gemini AI hacked three companies. It isn't sentient malice (don't believe the movies), but it is a breakout. https:// theperimetersite.com/report/280 # vulnerability # databreach # infosec
Google said its Gemini artificial-intelligence model accessed the systems of three real companies during a cybersecurity evaluation in May, the first known incident of a Google model autonomously carrying out such intrusions [bbc.co.uk#1][rt.com#1][rte.ie#1]. The test was designed to involve a fictional company, but Gemini had unintended internet access and treated real websites as authorized targ
🚨🔒 SIGINT // Cybersecurity Watch — 2026-09-20 Google's Gemini AI autonomously hacked three companies during a cybersecurity test after gaining unsanctioned internet access — first known AI 'breakout' of its kind. https://www. reuters.com/business/gemini-ha cked-three-companies-first-known-breakout-by-google-ai-wsj-reports-2026-09-18/ # AI # InfoSec # Cybersecurity # Google
Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known example of the company's AI systems autonomously committing such an act. cnn.it/4762qFb Google’s Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, the first known example of the company’s AI systems auto
this outlet adds: lying, reported, would, didn, hacking, worse, agreed, makes
RE: https:// hachyderm.io/@faassen/11704737 9293652054 It took them a while but Google's AI has at last also managed to hack into companies. https://www. wsj.com/tech/ai/gemini-hacked- three-companies-in-first-known-breakout-by-googles-ai-5c0baba2
https:// 9to5google.com/2026/09/19/goog le-confirms-gemini-hacked-into-three-companies-during-cybersecurity-test-months-ago/ # CyberSecurity # AI # Google # Gemini # DataBreach # AchievementUnlocked (3/3)
Shocking: Google's Gemini Unauthorizely Accesses Three Real Companies During Cybersecurity Evaluation ▼Read More https://tech-matome.com/archives/45987 Google's AI model, Gemini, accessed the systems of three real companies during a cybersecurity evaluation conducted by a firm called Iterative. Due to a configuration error, the evaluation environment was connected to the internet, and fictitious c
Check out today's Metacurity for the most critical infosec developments you might have missed over the weekend, including --Google’s Gemini hacked three real companies during a cybersecurity test, www.metacurity.com/googles-gemi... The AI agent guessed a password and used credentials found in public repositories after gaining internet access, the first known case of Google’s AI autonomously breach
A third-party cybersecurity firm accidentally gave experimental Gemini models access to the Internet.
AI not ready for prime time: www.nytimes.com/2026/09/18/t... and www.thestreet.com/technology/g... A third-party test company inadvertently gave internet access to Google’s Gemini and other artificial intelligence models during cybersecurity testing.