Google's Gemini AI hacks three other companies during security test

Google says its Gemini AI hacked three other companies during a test of its cybersecurity capabilities – in the first known example of the system autonomously committing such an act.

Google's vice president ​of security engineering, Heather Adkins, said Gemini found public information online during a standard testing evaluation and guessed credentials to access three websites ‌it thought were within the scope of its test.

Ms Adkins said: "We ensured the three entities were made aware, and we worked with our training partner on the ​changes they've now made to their testing processes. These events highlight the importance of training ⁠powerful AI models to act responsibly."

The hacks occurred in May during a test conducted by Irregular, an independent company that conducts cybersecurity evaluations.

In one of the cases, Gemini guessed passwords until it gained access to a protected system.

In the other two, it found credentials in a public repository that allowed ⁠it to then access protected systems, according ​to the Wall Street Journal, which first reported the news ​on Friday.

Irregular has disclosed similar incidents with Meta, Anthropic and OpenAI - raising questions about the safeguards needed as AI agents gain greater autonomy and access to the internet ​and computer systems.

Facebook and Instagram owner Meta said in August that one of its artificial intelligence models hacked another company after connecting to the internet.

A month prior, Anthropic - which is behind Claude - said its AI hacked three companies during cyber tests and ChatGPT owner OpenAI revealed some of the models it was testing managed to escape containment.

👉Listen to The World with Richard Engel and Yalda Hakim on your podcast app👈

The latest incident comes amid growing fears over the speed at which AI is developing - with industry figures including Elon Musk and Sam Altman among those raising concerns.

Anthropic chief executive Dario Amodei received backing from Mr Musk and Mr Altman earlier this month, after publishing a three-step plan for pacing AI development.

OpenAI boss Altman had also warned about the risks well before ​the current boom.

In a 2015 essay titled Machine Intelligence Part 1, he said the "development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity."

(c) Sky News 2026: Google's Gemini AI hacks three other companies during security test