Google: Attackers fired 100,000+ prompts in attempt to clone Gemini

Google says its Gemini AI has faced large-scale attempts by commercial actors to extract and replicate its underlying technology.

Google has disclosed that its flagship AI chatbot, Gemini, has been repeatedly targeted by “commercially motivated” actors attempting to clone it through massive prompting campaigns. In a report published Thursday, the company said one operation queried Gemini more than 100,000 times in what it described as “distillation attacks,” aimed at extracting the model’s logic and inner workings to build rival systems.

Google said the activity amounts to intellectual property theft, with most attacks believed to originate from private companies or researchers worldwide. “We’re going to be the canary in the coal mine for far more incidents,” said John Hultquist of Google’s Threat Intelligence Group, warning that smaller and custom AI tools are also vulnerable to similar model-extraction efforts.

READ MORE AT NBC NEWS.