Anthropic Accuses Alibaba of Massive AI Model Extraction Campaign
News Desk
25 June 2026, 14:01
Artificial intelligence competition between the United States and China has entered a new phase, with Anthropic accusing Chinese technology giant Alibaba of carrying out what it describes as the largest known attempt to extract capabilities from its Claude AI models.
According to a letter sent to senior members of the U.S. Senate Banking Committee and reviewed by Reuters, Anthropic alleges that operators affiliated with Alibaba and its AI division, Qwen, conducted a large-scale effort to harvest knowledge and capabilities from Claude between April and June this year.
The allegations add to growing concerns among U.S. policymakers and AI companies about intellectual property protection, model security, and the increasingly intense race to develop advanced artificial intelligence systems.
What Anthropic Is Alleging
Anthropic claims the operation relied on a technique known as "distillation," a process in which a less capable AI model is trained using the outputs generated by a more advanced system.
While distillation is commonly used within organizations to create smaller and more efficient models, AI companies have increasingly raised concerns about competitors using the technique to replicate proprietary capabilities without developing them independently.
According to Anthropic, the alleged campaign generated more than 28.8 million interactions with Claude through nearly 25,000 fraudulent accounts over six weeks. The company described the activity as the largest known extraction effort it has encountered.
Anthropic believes the operation was designed to accelerate the development of AI systems capable of matching features found in its advanced Mythos Preview models. Alibaba has not publicly responded to the allegations.
Growing Concerns Over AI Intellectual Property
The accusations come amid broader concerns about how leading AI models are being replicated and reverse-engineered. As AI systems become more powerful and expensive to develop, companies are increasingly treating model capabilities as strategic intellectual property.
Training frontier AI systems requires massive investments in computing infrastructure, talent, and data. As a result, firms have become more aggressive in protecting their technology from unauthorized copying or replication.
Anthropic previously disclosed similar concerns in February when it alleged that several Chinese AI companies, including DeepSeek, Moonshot AI, and MiniMax, had conducted campaigns aimed at extracting capabilities from Claude.
According to the company, those earlier operations ranged from hundreds of thousands to millions of interactions. The newly reported Alibaba case is significantly larger in scale.
AI Security Becomes a Geopolitical Issue
The dispute reflects how AI development is increasingly becoming intertwined with national security and geopolitical competition. In April, the White House accused China of systematically targeting intellectual property from leading American AI companies. U.S. officials have repeatedly warned that advanced AI technologies could have significant military, intelligence, and economic implications.
Anthropic has voiced support for government efforts to counter such activities, including increased cooperation between AI companies and federal agencies through threat intelligence sharing and security coordination.
The company's June letter was addressed to Senators Tim Scott and Elizabeth Warren ahead of a Senate hearing focused on artificial intelligence and national security.
A Complicated Policy Environment
The allegations emerge during a period of escalating scrutiny surrounding both Chinese and American AI firms.
Earlier this month, Alibaba was added to a Pentagon list of companies allegedly linked to China's military-industrial ecosystem. The company is currently challenging that designation through legal action.
At the same time, Anthropic itself has faced regulatory challenges. Just days after the company sent its letter to lawmakers, the U.S. Commerce Department imposed restrictions on Anthropic's advanced Mythos and Fable models due to concerns they could be accessed by foreign military or intelligence users. Those restrictions ultimately led Anthropic to disable access to the models globally.
The situation highlights the increasingly complex environment facing AI companies as governments attempt to balance innovation, national security, and international competition.
The Bigger Picture
Whether Anthropic's allegations against Alibaba are ultimately proven or disputed, the incident underscores a growing challenge for the AI industry. As frontier models become more powerful, protecting their capabilities is emerging as a major concern alongside issues such as safety, regulation, and infrastructure.
The case also illustrates how AI competition is evolving beyond hardware and computing resources. Increasingly, companies are battling over model capabilities, training techniques, and intellectual property that could shape the next generation of AI systems.
With governments becoming more involved and geopolitical tensions influencing technology policy, disputes over AI model extraction are likely to become more common as the global race for artificial intelligence leadership accelerates.
