The brewing “Cold War” between US and Chinese AI giants just hit a boiling point. On Tuesday, Anthropic, the San Francisco-based creator of Claude AI, officially accused three major Chinese AI labs, DeepSeek, MiniMax, and Moonshot AI, of “industrial-scale” data theft.

The allegations suggest a sophisticated campaign to “siphon” the intelligence of Claude to bolster Chinese models, just as DeepSeek prepares to launch its highly anticipated R2 model.

The Allegation: 16 Million Exchanges and “Model Distillation”

In a detailed blog post, Anthropic alleged that these companies bypassed safety protocols to “cheat” their way to better performance. The company provided concrete divs to back its claims:

Say Goodbye to Debt and Become Rich, Just Carry Them in Your Wallet

say goodbye to debt and become rich, just carry them in your wallet...

February 24, 2026

5:55 am

If You Find Moles or Skin Tags on Your Body, Read About This Remedy. Genius!

if you find moles or skin tags on your body, read about this remedy. genius!...

February 24, 2026

5:51 am

I weighed 332 lbs, and now 109! My diet is very simple trick. 1/2 Cup Of This (Before Bed)

i weighed 332 lbs, and now 109! my diet is very simple trick. 1/2 cup of this (before bed)...

February 24, 2026

5:57 am

This Simple Trick Removes All Parasites From Your Body!

this simple trick removes all parasites from your body!...

February 24, 2026

5:42 am

• 24,000 Fraudulent Accounts: Created specifically to query Claude.
• 16 Million Exchanges: The volume of data extracted through these conversations.
• Millions of Tokens: Conversations focused specifically on coding, reasoning, and tool use.

Anthropic refers to this process as distillation, training a smaller or newer model on the outputs of a superior one. While distillation is a common internal practice for companies like OpenAI and Anthropic to optimize their own models, Anthropic argues that doing so without permission is “illicit” and strips away vital safety guardrails.

“Models built through illicit distillation are unlikely to retain those safeguards, meaning that dangerous capabilities can proliferate with many protections stripped out entirely,” Anthropic stated.

The “Hypocrisy” Debate: Internet Reacts with Memes and Malice

Despite the gravity of the accusations, the court of public opinion isn’t exactly siding with the US giant. Social media platforms like X (formerly Twitter) were quickly flooded with accusations of “gaslighting.”

Recent Posts

Do This Every Night and the Fungus Will Disappear in 5 Days

do this every night and the fungus will disappear in 5 days...

February 24, 2026

5:59 am

This product is putting plastic surgeons out of work

this product is putting plastic surgeons out of work...

February 24, 2026

5:43 am

This method will instantly start hair growth

this method will instantly start hair growth...

February 24, 2026

5:51 am

Varicose veins will go away ! The easiest way!

varicose veins will go away ! the easiest way!...

February 24, 2026

5:45 am

Critics pointed to Anthropic’s own recent legal hurdles, including a $1.5 billion settlement involving authors who accused the company of using pirated books from torrent sites to train 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

Key voices from the internet:

• The Content Creator: “I spent years building a website… Claude parrotted it almost word-for-word. Spare us the gaslighting,” wrote one user.
• The Privacy Advocate: Others noted that the Chinese companies actually paid for API access, unlike the raw scraping methods often used by US companies on public websites.
• Elon Musk: The xAI boss didn’t hold back, calling Anthropic “smug” and “hypocritical,” adding, “Anthropic is guilty of stealing training data at massive scale… This is just a fact.”

Geopolitical Stakes: Chips, Exports, and the R2 Looming

This isn’t just a corporate spat; it’s a matter of national security. The timing of Anthropic’s attack coincides with two major events:

1. Export Control Scandals: Reports recently surfaced via Reuters suggesting DeepSeek may have used advanced Nvidia Blackwell chips—currently restricted by US export laws—at a data center in Inner Mongolia.
2. The DeepSeek R2 Launch: DeepSeek, which shocked the industry with its R1 model, is rumored to release DeepSeek R2 within the week. Early whispers suggest R2 could match or even exceed the capabilities of Claude 3.5 or GPT-4o.

Why This Matters for the AI Market

The core of the tension lies in the open-source vs. proprietary battle. Chinese models like DeepSeek are often open-source, allowing global developers to use their “weights” for free or at a low cost.

If Chinese models achieve parity with American models by using American data, the commercial “moat” around companies like Anthropic and OpenAI could evaporate. For the US, this isn’t just about “cheating”; it’s about losing the economic and strategic lead in the most important technology of the century.