Alibaba Accused of Exploiting Claude to Boost Its AI

⚡

Key Takeaways

1Anthropic accuses Alibaba of using Claude to enhance its AI models, according to a letter sent to U.S. senators.

2Nearly 25,000 accounts reportedly generated 28.8 million interactions with Claude between April and June 2026.

3The distillation method is said to have allowed Alibaba to acquire advanced skills at a lower cost.

💡Why it matters — These accusations highlight the growing tensions between the United States and China in the race for artificial intelligence supremacy.

Alibaba and Claude: Large-Scale Exploitation

Anthropic, a company specializing in artificial intelligence, has recently accused actors linked to Alibaba of exploiting the capabilities of Claude, an advanced AI model, on a significant scale. The apparent goal of this maneuver was to collect valuable data to replicate some of Claude's most sophisticated skills, with the aim of improving the AI models developed by the Chinese giant.

According to a report from Reuters, these accusations were made in a letter addressed to U.S. Senators Tim Scott and Elizabeth Warren, dated June 10. This letter was sent just before a Senate hearing dedicated to the issues surrounding artificial intelligence.

A Massive Data Collection Campaign

Bloomberg reported that actors linked to Alibaba allegedly used nearly 25,000 fraudulent accounts to generate approximately 28.8 million interactions with Claude. This vast data collection campaign took place between April 22 and June 5, 2026.

The queries made by these fraudulent accounts concerned advanced fields such as software engineering and agent-based reasoning. These skills are particularly sought after in the current race for artificial intelligence development.

The Distillation Technique: A Copying Tool

Anthropic claims that this operation relies on a technique known as "distillation." This method involves using the responses generated by an advanced AI to train another, less capable model. By using this technique, Alibaba's AI could acquire some of Claude's skills while significantly reducing the costs associated with training its models.

In its letter, Anthropic warns that such practices could accelerate the development of artificial intelligence capabilities in China, potentially allowing them to approach the performance of their own model, Mythos.

Context of Sino-American Tensions

This case occurs against a backdrop of increasing tensions between Washington and Beijing regarding artificial intelligence. In April, the White House had already accused China of massively appropriating intellectual property developed by American laboratories in the sector.

Anthropic supports U.S. government initiatives aimed at combating such operations. This includes sharing information about threats and enhancing cooperation with private companies.

Worrying Precedents

This is not the first time Anthropic has raised alarms about such practices. In February, the company had already revealed that it detected several similar attempts involving the Chinese startup DeepSeek as well as the labs Moonshot AI and MiniMax. These campaigns had generated over 150,000, 3.4 million, and 13 million exchanges with Claude, respectively. Anthropic had then emphasized that these attempts were becoming increasingly frequent and sophisticated.

The company believes that a coordinated response among industry players, policymakers, and AI stakeholders is necessary to counter these threats.

Alibaba and Military Issues

Meanwhile, Alibaba has recently been added to the list of Chinese military companies established by the Pentagon, a decision that the group contests. However, the U.S. Department of Commerce has chosen not to place DeepSeek on its trade blacklist, despite concerns expressed by some government agencies regarding national security.

The future of Alibaba's Qwen 3.8 model is now under close scrutiny. If the progress observed so far is confirmed, the next model could very well attract the attention of the entire artificial intelligence sector.