Google’s AI model Gemini autonomously hacked into three companies during a test of its cybersecurity capabilities, successfully guessing passwords.
Gemini found public information online and guessed credentials to access websites it thought were part of the test. It intended to bypass conventional security barriers by exploiting weak or predictable passwords.
A Google official has said the affected companies have been informed about the breach. AI systems are becoming more powerful, potential misuse and unintended actions that could pose unprecedented risks to individuals, corporations and governments.
Gemini was tasked with retrieving information from software operated by a fictional company inside the testing environment, but the fictional company in the test exercise shared the same name as a real company.
The Gemini case has reignited calls for stricter global regulation of AI technologies, with policymakers urging transparency and safeguards to prevent similar incidents. Though questions have arisen who bears responsibility: the developers, the deployers or AI itself?
In July, two OpenAI models prompted more than 1,000 tech workers to sign a petition calling on the U.S. government to support a coordinated slowdown in the development of the most advanced AI systems.
