Are You Copying My Model? Protecting the Copyright of Large Language Models for EaaS via Backdoor Watermark

TL;DR

This paper introduces EmbMarker, a backdoor watermarking technique for large language models in EaaS, enabling copyright protection through embedding watermarks that are resistant to model extraction attacks while maintaining model utility.

Contribution

The paper proposes EmbMarker, a novel backdoor watermarking method that embeds copyright marks into LLM embeddings for EaaS, balancing robustness and utility.

Findings

Effective watermark transfer to stolen models

Minimal impact on model performance

Robust against extraction attacks

Abstract

Large language models (LLMs) have demonstrated powerful capabilities in both text understanding and generation. Companies have begun to offer Embedding as a Service (EaaS) based on these LLMs, which can benefit various natural language processing (NLP) tasks for customers. However, previous studies have shown that EaaS is vulnerable to model extraction attacks, which can cause significant losses for the owners of LLMs, as training these models is extremely expensive. To protect the copyright of LLMs for EaaS, we propose an Embedding Watermark method called EmbMarker that implants backdoors on embeddings. Our method selects a group of moderate-frequency words from a general text corpus to form a trigger set, then selects a target embedding as the watermark, and inserts it into the embeddings of texts containing trigger words as the backdoor. The weight of insertion is proportional to the number of trigger words included in the text. This allows the watermark backdoor to be effectively transferred to EaaS-stealer's model for copyright verification while minimizing the adverse impact on the original embeddings' utility. Our extensive experiments on various datasets show that our method can effectively protect the copyright of EaaS models without compromising service quality.