OpenAI's Astra becomes the main target for Chinese AI labs after Anthropic tightens its defenses

Chinese AI startups DeepSeek and Moonshot allegedly used data generated by OpenAI's GPT-4o, referred to in some reports as GPT-6 Astra, to train their own large language models. This practice, often called model distillation, involves training a smaller model on the outputs of a more advanced one to improve performance efficiency. The report highlights ongoing questions regarding the use of proprietary AI outputs to develop competing technology and the potential implications for intellectual property and safety policies.