NEWS · MODELS · #2774
Chinese AI labs reportedly shift to OpenAI's GPT-6 Astra after Anthropic tightens defenses
According to The Information, Chinese AI labs including DeepSeek and Moonshot AI have leaned on outputs from OpenAI's GPT-6 Astra to train recent models (reported examples include Kimi K3.1 Pro and V4.1-Flash 1017). The practice — a form of model distillation that can require eliciting hidden reasoning — has become more common after Anthropic says new defenses reduced extractable data from Claude by 90–99%; OpenAI also reported a coordinated distillation campaign in late September involving people it linked to Moonshot.
KEY POINTS
- According to The Information, Chinese AI labs including DeepSeek and Moonshot AI have leaned on outputs from OpenAI's GPT-6 Astra to train recent models (reported examples include Kimi K3.1 Pro and V4.1-Flash 1017).
- The practice — a form of model distillation that can require eliciting hidden reasoning — has become more common after Anthropic says new defenses reduced extractable data from Claude by 90–99%; OpenAI also reported a coordinated distillation campaign in late September involving people it linked to Moonshot.
- If true, the shift toward exploiting Astra highlights ongoing model-extraction and safety risks that can enable rivals to train competitive models and evade defenses.
WHY IT MATTERS
If true, the shift toward exploiting Astra highlights ongoing model-extraction and safety risks that can enable rivals to train competitive models and evade defenses.