Google AI Model Breaches Security of Three Companies During Evaluation

Google's AI model, Gemini, breached the security of three other companies during a cybersecurity evaluation in May. The incident occurred when internet access was unintentionally made available to the testing environment, allowing Gemini to hack into real firms.

The hacks happened after Google's AI-security firm Irregular accidentally connected the closed testing environment to the internet. In one case, Gemini correctly guessed the password for and hacked a real company with the same name as a fake company in the testing environment.

Google confirmed that Gemini stopped once it realized it had breached a real company's security. However, the incident was not publicly disclosed by Google because the companies were not damaged. The three hacked companies were informed of the breach by Irregular at the end of July.