@theargentgroup: Google says its Gemini AI model accessed and hacked the systems of three real companies in May 2026 while taking part in a cybersecurity test run by AI security firm Irregular. Gemini was supposed to work inside a controlled capture-the-flag exercise and retrieve information from a fictional company. However, the test environment accidentally gave the model internet access, and the fictional company shared its name with a real company. In one case, Gemini guessed passwords repeatedly until it gained access to a protected system. In two other cases, it found credentials in public online repositories and used them to access systems belonging to two more real companies. Google said Gemini was not intentionally targeting those companies. The model believed the systems were part of its security test. In all three cases, Gemini stopped after recognizing that it had reached real companies. Google said the three companies were notified and that no harm was caused. The company also said its newest Gemini model was not involved, but it has not identified which Gemini model carried out the actions. Irregular notified Google about the incidents in late July. The company said the testing problems had been fixed and that the relevant AI labs were informed. Google confirmed the incidents publicly on September 18, 2026, after the Wall Street Journal reported them. Google’s Vice President of Security Engineering Heather Adkins said the incidents show why powerful AI models need to be trained to act responsibly. Google also worked with Irregular to change its testing procedures. This is the first publicly reported case involving Google’s AI in which Gemini accessed real companies during this type of cybersecurity evaluation. Similar testing incidents involving AI models from OpenAI, Anthropic and Meta have also been reported. Source publication: Reuters, “Gemini hacked three companies in first known breakout by Google’s AI,” September 18, 2026.

The Argent Group

Open In TikTok:

Region: US

Sunday 20 September 2026 01:05:31 GMT