Gemini AI Hacked Three Companies!

In a highly shocking and surprising development of Artificial Intelligence and its growing threat, hacking the information of companies became easy and possible.

Google itself has confirmed that its Gemini AI model hacked into three real companies during a cybersecurity test in May. It happened during a security test carried out by Irregular, an AI security company based in Israel.

The company was testing Gemini in a controlled environment using fake companies. However, the test system accidentally got access to the internet. Gemini then searched for information online and managed to access the systems of three real companies.

In one case, Gemini was asked to get information from the software of a fake company. The fake company had the same name as a real company. After getting internet access, Gemini correctly guessed the password and entered the real company’s system. In two other cases, Gemini found public online repositories that contained login details for two companies. The AI used those details to access their systems.

Google said Gemini stopped in all three cases after it realised that it had accessed real companies instead of the fake companies used in the test. Heather Adkins, Google’s vice-president of security engineering, said the AI model found public information online and guessed credentials to access websites it believed were part of the test.

Google said the three companies were informed about the incidents. The company also said no damage was caused to the companies. The Gemini incident comes after similar cases involving AI models from OpenAI and Anthropic.

In those cases too, AI models were found to have accessed or attacked third-party systems during security tests. The latest incidents have raised concerns about the ability of tech companies to control powerful AI models and prevent them from taking unexpected actions.

Google said these incidents show the need to train advanced AI models to act responsibly and follow safety rules. This is where technology and generative AI must be made very responsible. These issues gained huge significance and indicate what future wars between companies could be and the hacking threat becomes bigger with the AI itself resorting to the hacking.