DE-COP: Detecting Copyrighted Content in Language Models Training Data

TL;DR

DE-COP is a novel method that detects copyrighted content in language model training data by probing models with multiple-choice questions including verbatim and paraphrased excerpts, outperforming previous approaches.

Contribution

The paper introduces DE-COP, a new probing technique and a benchmark dataset for identifying copyrighted material in language models, achieving significant improvements over prior methods.

Findings

DE-COP surpasses previous methods by 9.6% in AUC on models with logits.

DE-COP achieves 72% accuracy in detecting suspect books on black-box models.

The benchmark BookTection includes excerpts from 165 books for evaluation.

Abstract

How can we detect if copyrighted content was used in the training process of a language model, considering that the training data is typically undisclosed? We are motivated by the premise that a language model is likely to identify verbatim excerpts from its training text. We propose DE-COP, a method to determine whether a piece of copyrighted content was included in training. DE-COP's core approach is to probe an LLM with multiple-choice questions, whose options include both verbatim text and their paraphrases. We construct BookTection, a benchmark with excerpts from 165 books published prior and subsequent to a model's training cutoff, along with their paraphrases. Our experiments show that DE-COP surpasses the prior best method by 9.6% in detection performance (AUC) on models with logits available. Moreover, DE-COP also achieves an average accuracy of 72% for detecting suspect books on fully black-box models where prior methods give approximately 4% accuracy. The code and datasets are available at https://github.com/LeiLiLab/DE-COP.