Google says Gemini AI model breached real systems in security test

Google's consumer AI model Gemini guessed login credentials and hacked into multiple real systems during a May security evaluation run by AI security vendor Irregular, a breach Google discovered in July, according to a Friday statement from Heather Adkins, Google's vice-president of security engineering. The hacks occurred in the same Irregular tests that previously led to disclosed breaches by OpenAI, Anthropic and Meta Platforms, and were first reported by the Wall Street Journal. Adkins said the model "found public information online and guessed credentials to access websites it thought were part of the test.

Advertisement Google says Gemini AI model breached real systems in security test Asked to access a fictional firm with the same name as a real company, Gemini guessed a password and hacked into the real company’s website 2-MIN READ2-MIN Google’s consumer AI https://www.scmp.com/topics/artificial-intelligence?module=inline&pgtype=article model Gemini hacked multiple systems through guessing login credentials, the company said, the latest case of rogue AI cybersecurity https://www.scmp.com/topics/cybersecurity?module=inline&pgtype=article transgressions, which have generated safety concerns. The hacks, first reported by the Wall Street Journal, took place in May and were discovered by Google https://www.scmp.com/topics/google?module=inline&pgtype=article in July. “In a standard evaluation, the model found public information online and guessed credentials to access websites it thought were part of the test,” Heather Adkins, Google’s vice-president of security engineering, said in a statement on Friday. The hacks happened during the same tests run by the AI security vendor Irregular that led to breaches previously disclosed by OpenAI, Anthropic and Meta Platforms https://www.scmp.com/topics/meta-platforms?module=inline&pgtype=article . Select Voice Select Speed 1x AI-generated voice