The review of the updated R1 AI model of DeepSeek is more stringent.

DeepSeek's Updated R1 AI Model Undergoes Stricter Censorship

The latest AI model of Chinese AI startup DeepSeek, an updated version of the company's R1 inference model, has achieved impressive results in benchmarks for programming, mathematics, and general knowledge, almost surpassing OpenAI's flagship model o3. However, the upgraded R1 may also be less willing to answer controversial questions, especially topics considered controversial by the Chinese government. This conclusion is drawn from tests conducted by the developers behind SpeechMap, a platform that compares how different models handle sensitive and controversial topics. According to them, R1-0528 censors responses to topics such as internment camps in the Xinjiang region. Although the model sometimes criticizes certain aspects of Chinese government policies, it usually gives the official Chinese government stance when directly asked relevant questions.

—— Techcrunch

via Windflag Reference Express - Telegram Channel