Anthropic: Chinese AI firms created 24,000 fraudulent accounts for distillation attacks
Mashable
February 23, 2026
AI-Generated Deep Dive Summary
Anthropic has accused three Chinese AI companies—DeepSeek, Moonshot, and MiniMax—of launching large-scale distillation attacks to reverse-engineer its Claude language model. These companies created over 24,000 fraudulent accounts to generate more than 16 million interactions with Claude, violating Anthropic’s terms of service. Distillation attacks involve repeatedly running similar prompts to replicate a model's capabilities, a legitimate technique often used for creating smaller, cost-effective versions of large language models (LLMs). However, in this case, the practice was exploited to steal technology and gain a competitive edge.
The issue has sparked debates over intellectual property rights and copyright laws in AI development. While Anthropic and other major AI companies claim they have the right to train their models on copyrighted materials without permission or payment, critics argue that such practices create ethical dilemmas. For instance, former President Donald Trump commented on the situation, suggesting that China’s approach to IP theft provides an unfair advantage in AI development.
AI companies are currently investing billions of dollars in AI infrastructure, and distillation attacks like these could undermine their competitive edge. Anthropic has called for increased cooperation between industry players, government agencies
Verticals
tech
Originally published on Mashable on 2/23/2026