DeepSeek Accused of Training AI Model with Google's Gemini Data
Summary
Researchers speculate that DeepSeek, a Chinese AI lab, may have trained its latest model using outputs from Google's Gemini AI, raising concerns about potential misuse of proprietary data and highlighting the challenges of preventing distillation in the rapidly evolving AI landscape.
Key Points
- DeepSeek may have used outputs from Google's Gemini AI model to train its latest R1 reasoning model
- Evidence suggests DeepSeek's model shares similarities with Gemini, such as preferred words and expressions, and thought process traces
- AI companies are ramping up security measures to prevent distillation, where models are trained on outputs from other AI systems