Extracting Memorized Training Data via Decomposition
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866913526379970560 |
|---|---|
| author | Su, Ellen Vellore, Anu Chang, Amy Mura, Raffaele Nelson, Blaine Kassianik, Paul Karbasi, Amin |
| author_facet | Su, Ellen Vellore, Anu Chang, Amy Mura, Raffaele Nelson, Blaine Kassianik, Paul Karbasi, Amin |
| contents | The widespread use of Large Language Models (LLMs) in society creates new information security challenges for developers, organizations, and end-users alike. LLMs are trained on large volumes of data, and their susceptibility to reveal the exact contents of the source training datasets poses security and safety risks. Although current alignment procedures restrict common risky behaviors, they do not completely prevent LLMs from leaking data. Prior work demonstrated that LLMs may be tricked into divulging training data by using out-of-distribution queries or adversarial techniques. In this paper, we demonstrate a simple, query-based decompositional method to extract news articles from two frontier LLMs. We use instruction decomposition techniques to incrementally extract fragments of training data. Out of 3723 New York Times articles, we extract at least one verbatim sentence from 73 articles, and over 20% of verbatim sentences from 6 articles. Our analysis demonstrates that this method successfully induces the LLM to generate texts that are reliable reproductions of news articles, meaning that they likely originate from the source training dataset. This method is simple, generalizable, and does not fine-tune or change the production model. If replicable at scale, this training data extraction methodology could expose new LLM security and safety vulnerabilities, including privacy risks and unauthorized data leaks. These implications require careful consideration from model development to its end-use. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_12367 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Extracting Memorized Training Data via Decomposition Su, Ellen Vellore, Anu Chang, Amy Mura, Raffaele Nelson, Blaine Kassianik, Paul Karbasi, Amin Machine Learning Artificial Intelligence Cryptography and Security The widespread use of Large Language Models (LLMs) in society creates new information security challenges for developers, organizations, and end-users alike. LLMs are trained on large volumes of data, and their susceptibility to reveal the exact contents of the source training datasets poses security and safety risks. Although current alignment procedures restrict common risky behaviors, they do not completely prevent LLMs from leaking data. Prior work demonstrated that LLMs may be tricked into divulging training data by using out-of-distribution queries or adversarial techniques. In this paper, we demonstrate a simple, query-based decompositional method to extract news articles from two frontier LLMs. We use instruction decomposition techniques to incrementally extract fragments of training data. Out of 3723 New York Times articles, we extract at least one verbatim sentence from 73 articles, and over 20% of verbatim sentences from 6 articles. Our analysis demonstrates that this method successfully induces the LLM to generate texts that are reliable reproductions of news articles, meaning that they likely originate from the source training dataset. This method is simple, generalizable, and does not fine-tune or change the production model. If replicable at scale, this training data extraction methodology could expose new LLM security and safety vulnerabilities, including privacy risks and unauthorized data leaks. These implications require careful consideration from model development to its end-use. |
| title | Extracting Memorized Training Data via Decomposition |
| topic | Machine Learning Artificial Intelligence Cryptography and Security |
| url | https://arxiv.org/abs/2409.12367 |