US AI Giant Accuses Chinese Rivals of Mass Data Theft

In a shocking revelation, US AI giant Anthropic has accused three Chinese firms of engaging in industrial-scale intellectual property theft.

TL;DR
• US AI giant Anthropic accuses Chinese rivals of mass data theft.
• The companies allegedly used a technique called distillation to extract capabilities from Anthropic's Claude chatbot.
• The practice poses national security risks and undermines US dominance in AI.

Summary
Anthropic, a U.S. AI firm, exposed a coordinated effort by three Chinese companies—DeepSeek, Moonshot AI, and MiniMax—to harvest its Claude chatbot’s capabilities through a method called distillation. The operation involved roughly 16 million exchanges and 24,000 fake accounts, enabling the rivals to replicate advanced features at a fraction of the development cost.
• Deploy rigorous monitoring of API usage to detect abnormal interaction patterns.
• Implement strict access controls and audit trails for third‑party integrations.
• Establish cross‑industry coalitions to share threat intelligence and coordinate regulatory responses.
• Invest in transparent model documentation to trace provenance and guardrail integrity.
• Allocate resources for continuous bias and safety audits of downstream models.

The distillation technique, common in AI development, was weaponized to bypass export controls and national security safeguards. MiniMax led the effort with over 13 million exchanges, focusing on coding, agentic reasoning, and tool use—areas where Claude excels. The firms routed traffic through proxy services and fraudulent accounts to evade Anthropic’s China‑specific commercial ban.

Anthropic warned that models built via illicit distillation lack the safety guardrails designed to prevent misuse, such as restrictions on bioweapon development or cyberattacks. The company called for coordinated industry and government action, noting that no single entity can address the threat alone. Similar accusations were recently made by OpenAI, underscoring a broader pattern of intellectual property theft in the AI sector.

The incident highlights the need for long‑term risk assessment, workforce support for displaced roles, and robust oversight mechanisms. By embedding transparency, regular audits, and safety investments into AI development pipelines, firms can mitigate exploitation risks while preserving competitive advantage and protecting national security interests.

As the AI industry continues to evolve, it's essential to address the issue of industrial-scale intellectual property theft and ensure that companies prioritize security and safety above all else.