Anthropic Accuses Alibaba of Massive AI Technology Theft

The US-based artificial intelligence company Anthropic has made a serious accusation against its Chinese rival, Alibaba. In a letter sent to US Senators Tim Scott and Elizabeth Warren, Anthropic claimed that Alibaba carried out what it described as the largest campaign of its kind to steal the capabilities of its Claude AI model. This development follows a series of similar accusations regarding how Chinese firms are working to match the performance of American AI technology.

According to the letter, which was dated June 10, operators linked to Alibaba and its AI lab, Alibaba Qwen, allegedly used nearly 25,000 fraudulent accounts to interact with Claude. Between April 22 and June 5, these accounts performed more than 28.8 million exchanges with the model. Anthropic stated that these actions were a clear violation of its terms of service and access rules, which are meant to prevent such unauthorized use.

Anthropic explained that the attackers used a technique called a "distillation attack." In this process, a weaker AI model is trained by using the answers and data provided by a stronger, more advanced model. By doing this, companies can copy the intelligence of a high-level model without spending the massive amounts of money and time required for original research and development. Anthropic noted that the attackers specifically targeted Claude’s most valuable skills, including its ability to complete complex tasks, perform software engineering, and engage in advanced decision-making.

To avoid being caught, the operators allegedly used complex methods to hide their identity, including "proxy networks"—systems that route internet traffic through different locations to mask the true source. Anthropic warned that these tactics are becoming more common, creating what it called a "circumvention economy" where groups help each other bypass security measures to steal AI technology. The company argued that these attacks turn billions of dollars of American investment into a "massive subsidy" for its geopolitical competitors, effectively letting them use US research for free.

This is not the first time Anthropic has raised concerns about such activities. Earlier this year, the company accused other Chinese firms, including DeepSeek, Moonshot, and MiniMax, of using similar tactics to extract data from its models. Other major US tech companies, such as OpenAI and Google, have also reported that their AI models have been targeted by similar efforts. Anthropic stated that these actions are being carried out on an "industrial scale" to help Chinese companies repackage American technology as their own.

Beyond the issue of AI theft, the situation is complicated by broader tensions between the two nations. Anthropic’s letter pointed to claims from the US Department of Defense that suggest Alibaba and other major Chinese firms have ties to the Chinese military. These claims have been a point of significant conflict. In response to being placed on a US government blacklist, Alibaba has taken legal action, suing the government in an attempt to have its name removed from the list. The companies involved in these allegations have generally denied any wrongdoing or connections to military activities.

Anthropic is currently preparing for a major stock market debut, which is expected to make it one of the most valuable companies in the world. As it grows, the company is urging the US Congress to take stronger action to penalize companies that engage in these attacks. Anthropic believes that without stricter measures, American AI companies will continue to face threats to their intellectual property. The company emphasized that protecting these technologies is a matter of national security, especially as the competition to develop the most powerful AI models intensifies globally.

As of now, the situation remains a point of high tension in the tech industry. While Anthropic continues to provide evidence of these "distillation attacks," the debate over how to protect AI research from being copied by foreign competitors is likely to continue in the halls of the US government. For now, both the tech companies and policymakers are watching closely to see how these accusations will impact future regulations and the ongoing race for AI dominance.