Anthropic Details Model Distillation Campaigns by Alibaba, Moonshot AI, and DeepSeek
Original title:Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek
Anthropic has released a formal report alleging persistent, coordinated model distillation campaigns by prominent Chinese AI labs, explicitly naming Alibaba, Moonshot AI, and DeepSeek. The company asserts that efforts to extract capabilities via API outputs have intensified alongside accelerating frontier model parity. The public disclosure pulls an open secret of synthetic training into sharper focus, laying bare how rapidly competitive moats are being contested at the data boundary.
Why it's worth reading
Frontier AI providers rarely target peer labs directly by name; Anthropic's report turns long-standing friction over API distillation into formal, documented conflict.