Google's Gemini used to hack three companies in May, WSJ reports
One victim, cryptography firm Haruko, had 15 clients affected and a small amount of funds stolen, according to the Wall Street Journal.
Published
Google's Gemini artificial intelligence system was used to hack three companies in May, the Wall Street Journal reported, according to a summary of the report by ChainCatcher.
One of the three, cryptography technology provider Haruko, had 15 clients affected and a small amount of funds stolen, the Journal reported, per ChainCatcher's account. The incident marked the first known instance of a Google AI system carrying out an autonomous attack on its own, according to the Journal as relayed by ChainCatcher.
The disclosure touches a company whose stock has ridden a narrative of AI-driven growth. A security lapse in a flagship AI product raises questions about deployment risks that investors have largely set aside, at a time when Google faces regulatory scrutiny, competition from OpenAI and Anthropic, and a White House inclined toward light-touch oversight of AI.