ET AI Explainer: Why are US AI labs wary of 'distillation attacks' from Chinese peers

Anthropic and OpenAI have raised concerns about unauthorized AI distillation attacks, where competitors, notably Chinese firms like DeepSeek and Moonshot AI, use their models' outputs to build cheaper, faster systems. Anthropic claims to have reduced such data extraction by 90-99% on its Claude models, while OpenAI disrupted a large-scale coordinated extraction campaign linked to Moonshot AI.