Elon Musk has expressed his disapproval of Anthropic following their allegations against Chinese AI companies, including DeepSeek, for purportedly scraping vast amounts of information from Claude to develop their own AI systems. Musk’s activity on the social media platform X (formerly Twitter) has been notable, particularly in his critique of the company led by Dario Amodei, which he claims is unlawfully acquiring data to enhance its Claude AI model.
Claims by Anthropic Against Chinese AI Firms
Anthropic has relayed an extensive blog entry accusing three Chinese AI entities, DeepSeek, Moonshot AI, and MiniMax, of utilising Claude’s outputs to train their AI models. The firm asserts that these companies have established over 24,000 fake accounts, generating more than 16 million interactions with Claude in an effort to improve their technologies.
Musk’s Response to the Allegations
In direct response to Anthropic’s allegations, Musk shared a post on X on February 23, highlighting that Anthropic has itself been involved in data theft on a significant scale and has faced multi-billion dollar settlements due to this. He stated that such claims are factual and cannot be disputed.
ADVERTISEMENT
Evidence Presented by Musk
To back his assertions, Musk provided screenshots from X Community Notes, which pointed out that Anthropic had previously resolved a lawsuit amounting to $1.5 billion concerning Claude by allegedly infringing on copyright by using over 7 million books from shadow libraries for training purposes. Moreover, the company is currently facing a separate lawsuit for purportedly torrenting over 20,000 songs, which further cements accusations of using stolen creative material while reprimanding others for similar actions.
Anthropic’s Findings on Distillation Attacks
Anthropic made known the existence of extensive distillation attacks aimed at their models by DeepSeek, Moonshot AI, and MiniMax. The statement highlighted that these laboratories created over 24,000 fraudulent accounts leading to more than 16 million interactions with Claude. The objective was to siphon its capabilities to further their own models.
— Anthropic (@AnthropicAI) February 23, 2026
Musk’s Continued Criticism of Anthropic
Beyond making substantiated accusations against Anthropic, Musk has been resharing multiple posts on X that echo similar allegations, including calls from individuals advising users to be cautious about trusting the firm. These statements seem deliberate, suggesting that Anthropic should scrutinise its own practices prior to levelling accusations against others for misusing their data.
The Broader Context of Distillation Attacks
Nevertheless, distillation attacks are a prevalent issue confronting companies in the AI sector. This technique is used to train a lesser-performing model by leveraging the outputs of a more robust model. Major tech entities, including OpenAI and Google, have reported similar incidents of suspected data misuse. Consequently, these companies are also reinforcing measures to safeguard against unauthorised extraction of their AI models.
