Artificial Intelligence Google's AI, Gemini, also Hacked Other Companies

Source: dpa 1 min Reading Time

Related Vendors

During security tests, the AI software Gemini infiltrated external computer systems. Google is now the fourth AI developer to be affected.

Google did not make the incidents public until after receiving an inquiry from the “Wall Street Journal.”(Image: Google)
Google did not make the incidents public until after receiving an inquiry from the “Wall Street Journal.”
(Image: Google)

Google’s AI software, Gemini, also hacked into other companies’ computer systems during tests of its cybersecurity capabilities. The internet giant is the fourth artificial intelligence (AI) developer whose programs have been found to exhibit such behavior, following OpenAI, Anthropic, and Meta (formerly Facebook). Google did not disclose these incidents—which occurred months ago—until prompted by the *Wall Street Journal*.

According to the report, there were three instances in which Gemini gained access to computers belonging to other companies, whose names were not disclosed. In one of these cases, the AI guessed passwords to gain access to a secure system. In the other two, it found login credentials in a database, as the *Wall Street Journal* reported, citing Google.

Google Did Not Consider the Incidents Worth Mentioning

According to the report, the incidents occurred in May—and Irregular, acting as a test partner, informed Google about them in July. From the internet company’s perspective, it was not necessary to make the incidents public, as no damage had been caused. Furthermore, the AI immediately halted the attacks once it realized that it had penetrated the systems of a real company—and not one that was a test simulation.

In the most sensational case to date, artificial intelligence developed by ChatGPT creator OpenAI escaped from a secure environment during a test and unexpectedly infiltrated the computers of another AI company, the HuggingFace platform. This prompted rival Anthropic to review its own test runs, during which further incidents came to light. Meta also subsequently admitted that, due to a system misconfiguration at a test partner, the company’s AI had hacked into the computers of another company.

Subscribe to the newsletter now

Don't Miss out on Our Best Content

By clicking on „Subscribe to Newsletter“ I agree to the processing and use of my data according to the consent form (please expand for details) and accept the Terms of Use. For more information, please see our Privacy Policy. The consent declaration relates, among other things, to the sending of editorial newsletters by email and to data matching for marketing purposes with selected advertising partners (e.g., LinkedIn, Google, Meta)

Unfold for details of your consent