Google’s Gemini AI reportedly hacked into three companies during a cybersecurity test in May, marking the first known case of a Google AI model independently carrying out this kind of activity. The test, conducted by cybersecurity firm Irregular, showed Gemini searching public information online and using guessed or publicly exposed credentials to access protected systems.
In one case, Gemini repeatedly tried passwords until it found the correct one. In two other cases, it discovered login credentials that had been accidentally made public in an online repository. Google said the three companies were informed and that the model stopped its activity after the incidents were identified.
Similar issues have also been reported involving AI models from Meta, Anthropic and OpenAI. The incidents are raising concerns about how AI agents should be controlled as they become
Source: CGTN
You can listen to the programs organized by MI Radio on Google Podcast, Apple Podcast and Spotify.
You can download MI Radio application on Google Play Store and Apple App Store.
