Anthropic Accuses Alibaba of Stealing Claude AI Data via Fake Accounts

Anthropic, the artificial intelligence company behind the widely used Claude AI model, has filed allegations accusing China's Alibaba Group of using fraudulent accounts to illicitly extract data and capabilities from its AI system. The claim represents one of the sharpest public accusations between a leading Western AI firm and a Chinese technology giant over alleged intellectual property theft.

What Anthropic Alleges

According to Anthropic, Alibaba created fake accounts to gain access to Claude and systematically extracted data and capabilities in a manner designed to bypass the terms of service. This technique — sometimes called model extraction — is a form of reverse engineering that attempts to replicate the behaviour of an AI model without legitimate access to its underlying architecture or training data.

Why This Matters

Artificial intelligence has become a central battleground in the broader geopolitical competition between the US and China, with both governments investing heavily in domestic AI and restricting technology exports. Allegations of deliberate IP theft by state-linked firms add another flashpoint to that rivalry.

Alibaba's Position

Alibaba has separately been suing the US government after being placed on a US Department of Defense list of companies allegedly with ties to the Chinese military. The two sets of legal disputes highlight deepening friction between the world's two largest technology ecosystems.

The Broader AI Race

The dispute underscores the competitive pressure surrounding frontier AI models. The capabilities embedded in models like Claude represent years of research investment. Companies accused of extracting these capabilities without authorisation shortcut that investment, gaining a competitive advantage at others' expense.