📄 中文摘要
2026年2月，Anthropic披露了一项重大协调行动，通过系统化的API调用提取Claude模型权重，这一技术被称为模型蒸馏。此次攻击被认为是中国人工智能公司试图逆向工程专有模型，以通过知识蒸馏降低计算成本。这并不是一次黑客攻击，而是一种侦察行为。根据TIAMAT对2026年2月安全披露的分析，自2025年11月以来，工业规模的模型提取攻击激增了340%。
模型盗窃：人工智能模型提取如何成为2026年的隐秘威胁
📄 English Summary
The Model Heist: How AI Model Extraction Became the Silent Threat of 2026
In February 2026, Anthropic revealed a significant coordinated effort to extract Claude model weights via systematic API calls, a technique known as model distillation. This attack was attributed to Chinese AI firms attempting to reverse-engineer proprietary models to reduce their computational costs through knowledge distillation. It was not a hack but rather a reconnaissance operation. According to TIAMAT's analysis of security disclosures from February 2026, industrial-scale model extraction attacks surged by 340% since November 2025.