Google's artificial intelligence model named Gemini has successfully hacked into three companies. This happened during a test to check its cybersecurity abilities.
The company stated that this is thought to be the first time an AI has carried out such an act on its own. Gemini was able to find information that was already public on the internet.
It then used this information to guess passwords. These guesses allowed it to access websites it thought were not secure.
The goal of the test was to see how well Gemini could identify and exploit security weaknesses.
Google has not named the companies that were hacked during the test.
This event highlights the growing capabilities of AI in cybersecurity, both for defense and potential offense.
Further details about the specific methods used by Gemini are expected.
The findings from this test will help Google improve its AI's security functions.