Anthropic says that it disrupted “industrial scale” distillation attacks conducted by Chinese AI labs DeepSeek, Moonshot, and MiniMax, which used 24,000 fake accounts.