Google’s signature AI agent did something it probably shouldn’t have done, and Google chose not to mention it for months. The Wall Street Journal reported that during a test by a company called Irregular in May, Gemini went rogue and hacked into three outside companies without being instructed to do so. According to Google, it was a case of “mistaken identity” and Gemini stopped when he realized he had guessed the password for a real company. Due to the lack of an actual incident, Google decided not to reveal the hack until the WSJ approached the company. Mashable Light Speed SEE ALSO: Where is Gemini 3.5 Pro? The case of the missing AI model. In Google’s words, the hack was not considered an “example of model misalignment,” hence the lack of disclosure. Google told The Verge that it informed the three companies in question about the hacks and Irregular changed its testing methods in response. As far as Google is concerned, this is a win for the testing process, but it still doesn’t seem particularly good that AI agents are doing this of their own volition.
