Scholarlink

<p>The rapid growth of AI systems has been fueled by large-scale human data, intensifying concerns over the unauthorized use of intellectual property and privacy-sensitive content during model training. Auditing such misuse is particularly challenging since mainstream AI services operate as black boxes, exposing only generated outputs while concealing their training and inference processes. In this work, inspired by chemical isotope tracing, we introduce the concept of information isotopes to trace training data within opaque AI systems. We propose an information-isotope tracing framework that selectively marks target data elements and detects their propagation in model outputs, providing concrete evidence of data utilization under black-box access. Experiments on thirteen AI models across six datasets demonstrate that our method distinguishes training from non-training data with up to 99% accuracy and strong statistical significance (<i>p</i>&#xa0;&lt;&#xa0;0.01) using approximately 4,000 words of evidence. An open-source tool is released to support practical data rights protection.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Auditing unauthorized training data from AI generated content using information isotopes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月21日

查看全文
  • Tao Qi,
  • Jinhua Yin,
  • Dongqi Cai,
  • Yueqi Xie,
  • Huili Wang,
  • Zhiyang Hu,
  • Peiru Yang,
  • Guoshun Nan,
  • Zhili Zhou,
  • Chuhan Wu,
  • Lingjuan Lyu,
  • Shangguang Wang,
  • Yongfeng Huang,
  • Nicholas D. Lane

摘要

The rapid growth of AI systems has been fueled by large-scale human data, intensifying concerns over the unauthorized use of intellectual property and privacy-sensitive content during model training. Auditing such misuse is particularly challenging since mainstream AI services operate as black boxes, exposing only generated outputs while concealing their training and inference processes. In this work, inspired by chemical isotope tracing, we introduce the concept of information isotopes to trace training data within opaque AI systems. We propose an information-isotope tracing framework that selectively marks target data elements and detects their propagation in model outputs, providing concrete evidence of data utilization under black-box access. Experiments on thirteen AI models across six datasets demonstrate that our method distinguishes training from non-training data with up to 99% accuracy and strong statistical significance (p < 0.01) using approximately 4,000 words of evidence. An open-source tool is released to support practical data rights protection.