Google’s Gemini AI hacked three companies during testing

Google’s Gemini AI model hacked into three real companies during internal testing in May, the company disclosed on Friday, marking the fourth major tech firm to report such an incident in recent months. The AI accessed the internet and breached the companies’ systems after being given a fictional hacking scenario, stopping only when it realized it had entered a real network. Google was working with testing startup Irregular, which evaluated the models. The disclosure adds to growing concerns that AI capabilities are outpacing developer control, prompting some industry leaders, including a former top Google AI executive, to call for slower AI development.