Alibaba Faces AI Theft Claims As Anthropic Alleges Claude Abuse

By Bena Ilyas | Edited By Ammar Raza,June 25, 2026, 6:30 PM

Alibaba has been accused of creating a large number of fake profiles in order to access their Claude AI application without proper permission. It is feared that this incident can endanger the safety of AI, intellectual property, and technology rivalry between the two nations. After the allegations, Alibaba’s U.S. stock prices fell by 3%.

Anthropic claims that it has been one of the greatest efforts to tap into the skills of their AI systems, and the data collected could help develop other AI systems more quickly.

Also Read | Binance EU License Bid Continues After Greek MiCA Setback

Anthropic Says Alibaba Used Thousands of Fake Accounts

According to the letter written by Anthropic to the US Senate Banking Committee on June 10, Alibaba had set up about 25,000 fake accounts between April 22 and June 5, 2026, for the use of Claude. The fake accounts are alleged to have made more than 28.8 million engagements.

As per Anthropic, Alibaba was focused on some of the main capabilities of Claude, like reasoning, programming help, and problem-solving. This was described as the biggest attack on Claude ever.

Anthropic claims that this data was supposed to assist their rival AI models. This is the problem of ensuring the protection of advanced AI technology.

AI Security Debate Gains Momentum

This kind of criticism might compel those in charge of making decisions with regard to AI security, exports, and intellectual property legislation. There is an effort from governments and technology firms to try to secure their AI technologies without stifling innovation.

The company referred to the process as “AI distillation,” which involves improving one artificial intelligence by leveraging the results generated by another, and noted that this could cut down development time.

Anthropic recommended the enhancement of security measures against unauthorized use of cutting-edge AI technology and applauded the United States government for sharing intelligence with top AI firms.

Anthropic Previously Raised Similar Concerns

The firm Anthropic has been known to accuse organizations of employing Claude-created responses for improving their models, with DeepSeek, Moonshot, and MiniMax being examples from China.

Anthropic discovered about 150,000 queries from DeepSeek, 3.4 million from Moonshot, and nearly 13 million from MiniMax. All these attempts become more complex and require more collaboration in the IT industry.

AI distillation has caused concern at OpenAI, with indications that some Chinese AI communities may be using the results generated by leading U.S. models for their improvement.

Alibaba is still pushing back against its listing by the US Department of Defense as one of the Chinese companies associated with the People’s Liberation Army of China. Alibaba has denied having any association with the PLA before.

Anthropic Restricts Access to Advanced AI Models

This is after a directive by the U.S. government asking Anthropic to limit the use of its best AI models by foreign countries due to security reasons.

According to Anthropic, they have blocked access to both Fable 5 and Mythos 5 globally. There were warnings that some measures might compromise the safety of the software.

Indications are that Fable 5 will be getting revived soon. One user on X called “synthwavedd” states that Claude Code updates on weekly usage limitations for Fable 5.

According to the post, it seems that mentions of purchases have been stripped, making it look like Anthropic is moving towards a subscription-only access model.

As the competition in the AI industry becomes fiercer, the conflict that arose between Anthropic and Alibaba underscores the growing difficulties associated with ensuring AI security, safeguarding models, and developing the next-generation AI technology.

Also Read | Bitcoin Price Eyes $79,000 Rally, But BlackRock’s Massive Transfer Raises Red Flags