New Delhi: Anthropic has alleged that three Chinese artificial intelligence firms had abused its Claude chatbot to reinforce their own artificial intelligence systems.
The assertions have initiated a new conversation around the use of data, model training and hypocrisy in the global AI sector.
The scandal only escalated when xAI chief Elon Musk openly criticised the startup headed by Dario Amodei. Musk alleged that Anthropic was stealing data in bulk itself, a move that brings the conflict between market leaders in artificial intelligence at a time when the world is rapidly growing more competitive in the field.
Musk alleges data theft by Anthropic
Elon Musk posted on X that Anthropic was guilty of data theft on a large scale and that the company had made multi-billion-dollar settlements over claims of its data abuse. To substantiate his assertion, he provided screen shots of X Community Notes that mentioned a purported $1.5 billion lawsuit settlement involving the creation of Claude.
Anthropic is guilty of stealing training data at massive scale and has had to pay multi-billion dollar settlements for their theft. This is just a fact. https://t.co/EEtdsJQ1Op
— Elon Musk (@elonmusk) February 23, 2026
The statements by Musk were made soon after the publication of Anthropic accusing Chinese companies of the unfair extraction of outputs of its models. His remarks gave instant viral publicity and fuelled an already fiery online conversation on the way AI systems are trained.
Anthropic flags ‘Industrial-Scale Distillation Attacks’
Anthropic wrote in a blog post that three Chinese AI firms, DeepSeek, Moonshot AI and MiniMax, had made over 24,000 fake accounts and had made over 16 million interactions with Claude.
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax.
These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
— Anthropic (@AnthropicAI) February 23, 2026
The companies applied a method referred to as distillation as per the company. In this approach, the input of a more sophisticated system is used to train a smaller or incompetent model. Although the industry has a common use of distillation to make lighter and more cost-effective models, Anthropic claimed that it becomes problematic when performed unauthorised.
According to the company, this type of activity might enable foreign labs to deprive safety safeguards and, in fact, use advanced AI capabilities in military, intelligence, or surveillance systems. It declared it a dire case and that there is a small window of time to act as such campaigns become more and more sophisticated.
Online backlash and accusations of hypocrisy
The arguments by Anthropic caused a social media backlash. Critics claimed that big AI research labs, such as Anthropic, used large quantities of internet data that is publicly available, much of which is copyrighted.
The fact that one can learn with the help of outputs at an API access price did not make sense to some users, and the long-term practice of mass scraping of online content during pretraining has been justified as fair use by AI companies. Others cited what others termed a two-sided approach, claiming that AI companies are comfortable with the extensive use of data to their advantage, but this does not apply to other companies using the same techniques.
The controversy points to the increasing conflict in the AI ecosystem, where intellectual property, data rights, and national security issues are becoming more and more conflictual.
