Anthropic Accuses Alibaba, Moonshot AI and DeepSeek of Running Distillation Campaigns Against Its Models

American company Anthropic, the developer of the Claude models, released a report on Thursday detailing what it describes as persistent distillation campaigns launched by AI firms based in China. According to TechCrunch, the companies named include Alibaba, Moonshot AI and DeepSeek, and these attempts are said to have intensified as global competition in the AI sector has grown increasingly fierce.

What Distillation Is — and Why It's a Problem

Distillation is a practice in which a smaller or newer AI model is trained using outputs generated by another, more advanced model, with the goal of replicating its performance without bearing the enormous costs of training from scratch. While the technique has legitimate research applications, Anthropic claims that some companies have used it to systematically extract knowledge from Claude, in violation of the service's terms of use.

The report comes at a time when the AI industry is facing increasingly serious questions about intellectual property and how companies protect their massive investments in developing large-scale language models.

Growing Tensions Between the US and China Over AI

According to the cited source, Anthropic says it has identified activity patterns suggesting the automated, large-scale use of its APIs to generate data intended for training competing models. However, the company has not provided full technical details about the methodology used to reach these conclusions, which could fuel skepticism among some industry observers.

The situation reflects broader tensions between American and Chinese AI companies at a time when both sides are competing for technological supremacy. DeepSeek, in particular, drew international attention last year after releasing models with performance comparable to Western counterparts but at significantly lower costs, raising questions about how it was trained.

Neither Alibaba, Moonshot AI, nor DeepSeek has publicly responded to Anthropic's allegations so far, according to available information.