Hackers are engaging in “distillation attacks,” where they use legitimate access to steal AI technology. According to a new report from Google, hackers have used over 100,000 AI prompts to replicate the company’s Gemini AI model. The attacks seem to be coming from adversaries in countries like North Korea, Russia, and China.
Google identifies these attempts as “model extraction attacks,” where hackers flood an AI system with thousands of prompts to replicate its capabilities. This is not a threat to Google users but rather to service providers and model builders who could have their work stolen.
The war over AI models has intensified, with Chinese companies like ByteDance introducing advanced tools. Last year, a Chinese AI company called DeepSeek introduced a rival model that rivaled the world’s top technology. OpenAI accused DeepSeek of using existing technology in ways similar to Google’s report.
Google is warning others about these threats, saying it may be the first to face such attacks but many more are likely to follow.
Source: https://www.cnet.com/tech/services-and-software/hackers-are-trying-to-copy-gemini-via-thousands-of-ai-prompts-says-google