Anthropic has accused foreign AI labs of stealing data from its AI model Claude to build their own smaller and cheaper models for the customers. These attacks are called distillation where they push the AI model to share
its secrets, how it was built and other confidential details. “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 mentioned in its post on X.
And while Anthropic has raised these concerns from Chinese threat, xAI CEO Elon Musk has taken a dig at the AI company and accused them of stealing data and using public data for free, for which, Musk claims, Anthropic was penalised $1.5 billion.
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
Musk Vs Anthropic: The New AI Battle
Musk replied to the Anthropic posts about distillation attacks by citing community notes on X, and also mentioned that the AI company had to pay multi-billion dollar settlement for its alleged stealing of data to train its AI models. These cross accusations have become fairly common these days, and the AI arena is riddled with them since the lack of training data could tempt them into a dark path.
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 community note under Anthropic’s own post suggests the company had pirated over 7 million books from shadow libraries to train Claude, and also for torrenting 20,000+ songs. Musk is basically using the classic, “Look who’s talking,’ phrase to take a dig at Anthropic and suggested they should look within themselves first before accusing anyone else of stealing its data.
Having said that, these AI-gen attacks will surely become a concern for more companies in the months and years to come.
Google had recently sounded the bugle regarding these attacks. It has even come out with a detailed report which says some attackers tried to prompt Gemini into making a clone version of the AI model which can have a massive impact on how the AI arena operates over the next few years.
The company is calling it “distillation attacks,” which is designed to push the AI chatbot to share the confidential details about its model and how it functions in the background.
Anthropic isn’t the first AI company to make the news for stealing accusations and won’t be the last. And with the industry looking for a quick start in the arena, expect more data scraping to happen without your knowledge.
