China Copying AI Technology From US Companies? Here's What Report Says
Anthropic, OpenAI and Google have accused Chinese AI companies of using “distillation” to copy capabilities from their proprietary models. However, experts argue the claims may be overstated, noting that distillation captures model outputs rather than underlying code. The legal status remains unclear as Chinese AI firms continue to narrow the technology gap.
Leading American artificial intelligence developers including Anthropic, OpenAI, and Google have spent nearly two years accusing Chinese counterparts of unfairly copying their proprietary systems. These allegations emerged as a major flashpoint during discussions between Chinese leader Xi Jinping and U.S. President Donald Trump in Washington. While U.S. firms lead the global AI race, fast-rising Chinese startups are rapidly narrowing the technology gap, fueling fierce geopolitical and commercial rivalries.
At the center of the dispute is an intricate technical process known as "distillation," which Silicon Valley executives claim Chinese companies use to misappropriate leading U.S. capabilities. As per a report by Indian Express, many industry experts argue that these claims are overblown and that distillation alone does not fully account for the competitive power of Chinese models. Alibaba Unveils China’s ‘Most Powerful’ AI Chip With 3x Performance, Plans 10 Trillion-Parameter Model.
Understanding the AI Distillation Process
Originally developed in the early 2010s, distillation was designed as a legitimate research method to build more efficient systems by collecting data from an established model to run on cheaper hardware. Pioneer Geoffrey Hinton frequently describes the mechanism as a dynamic between a teacher model and a student model. However, critics argue that the practice has evolved into a tool for competitors to harvest data from proprietary systems.
In practice, a competing developer sets up multiple accounts on services like Anthropic's Claude or OpenAI's GPT to run specific tasks in coding or mathematics. The generated outputs are then funneled into training a separate system. Legal experts note that while some argue this practice violates trade secret laws, U.S. courts have not reached a definitive ruling, and copyright laws do not neatly apply since the general behavior of the system is copied rather than literal text.
The Debate Over Originality and Enforcement
Accusations have targeted several prominent Chinese entities, including DeepSeek and Alibaba, following surprise releases of highly efficient models that challenge Western market dominance. Nevertheless, technical specialists emphasize that distillation only provides access to superficial system outputs and characters rather than underlying proprietary computer code. Building a competitive model still requires substantial foundational infrastructure, funding, and engineering talent. AI Serious Risks: China Joins Tech Leaders, Warns About National Security Threats of Artificial Intelligence.
Furthermore, American AI giants facing accusations of intellectual property theft are themselves embroiled in major copyright lawsuits concerning how they trained their own systems on internet data. While companies can attempt to block suspicious accounts to curb unauthorized distillation, researchers maintain that completely stopping the practice remains virtually impossible without disrupting legitimate everyday users.
(The above story first appeared on LatestLY on Sep 27, 2026 02:49 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).