Frontier AI Rip-Off? Anthropic Points Finger

Anthropic’s accusation against Alibaba puts a sharp edge on the fight over who gets to use frontier AI and who gets to copy it.

Quick Take

Anthropic says operators tied to Alibaba’s Qwen AI lab used nearly 25,000 fraudulent accounts.
The company says those accounts made more than 28.8 million Claude exchanges in about six weeks.
Anthropic says the activity targeted software engineering and agentic reasoning.
The dispute centers on distillation, a method that can be legal inside a company but illicit when used on a rival model.

What Anthropic Says Happened

Anthropic says it identified a large extraction effort linked to Alibaba and its Qwen AI lab. The company says the campaign ran from April 22 through June 5 and used nearly 25,000 fraudulent accounts to make 28.8 million exchanges with Claude. Anthropic says the activity focused on the model’s most valuable skills, including coding and agentic reasoning, which are the kinds of features firms pay for in business use.^[1]^[6]

The company’s case rests on a familiar AI security pattern. Anthropic says distillation is a legitimate method when labs train their own smaller models on outputs from stronger ones, but it says outside operators crossed the line by using fake accounts and proxy-like access to harvest Claude responses at scale. Anthropic says it can trace such campaigns through IP correlation, request metadata, infrastructure clues, and partner reports.^[1]^[20]

Why Conservatives Should Care

This is not just a tech fight. It is another sign that U.S. companies are trying to protect American innovation from theft, while Chinese firms are accused of taking shortcuts instead of building their own systems. The wider issue also touches national security, since Anthropic has warned that illicitly distilled models can strip out safety guardrails and feed military, intelligence, and surveillance systems abroad.^[20]^[22]

For readers already fed up with weak borders, weak rules, and weak enforcement, the pattern will look familiar. American firms spend billions to build advanced tools, then face claims that foreign rivals can copy the results through scale, fraud, and loopholes. Anthropic says that is why it treated the matter as a policy issue, not just a terms-of-service problem.^[1]^[3]

What Is Known and What Is Not

The public record supports Anthropic’s broad claim that it detected industrial-scale distillation attempts and that it attributes them with high confidence to specific labs using technical signals.^[20] The public record does not, however, include the raw forensic logs, account records, or technical proof needed for outside observers to independently verify every detail. That gap matters because Alibaba has not publicly released a detailed rebuttal tied to the specific allegation.^[1]^[2]^[3]

Anthropic accused Alibaba-linked operators of the largest distillation attack yet…

~25,000 fake accounts + 28.8 million interactions with Claude (April–June 2026) to steal advanced reasoning, coding & agentic capabilities.

This hits the core global issue.

Why does the…

— SunDeep Mehra (@SundeepMehra7) June 25, 2026

For now, the dispute sits in the same category as other recent AI theft claims: serious, detailed, and politically charged, but still driven mainly by company disclosures and media reporting. Reuters says Anthropic’s accusation is based on a letter it sent to lawmakers and White House officials, while other reports repeat the same account of fake accounts, millions of exchanges, and targeted capability extraction.^[6]^[8]^[9]