Measuring Copyright Risks of Large Language Model via Partial Information Probing
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914954385293312 |
|---|---|
| author | Zhao, Weijie Shao, Huajie Xu, Zhaozhuo Duan, Suzhen Zhang, Denghui |
| author_facet | Zhao, Weijie Shao, Huajie Xu, Zhaozhuo Duan, Suzhen Zhang, Denghui |
| contents | Exploring the data sources used to train Large Language Models (LLMs) is a crucial direction in investigating potential copyright infringement by these models. While this approach can identify the possible use of copyrighted materials in training data, it does not directly measure infringing risks. Recent research has shifted towards testing whether LLMs can directly output copyrighted content. Addressing this direction, we investigate and assess LLMs' capacity to generate infringing content by providing them with partial information from copyrighted materials, and try to use iterative prompting to get LLMs to generate more infringing content. Specifically, we input a portion of a copyrighted text into LLMs, prompt them to complete it, and then analyze the overlap between the generated content and the original copyrighted material. Our findings demonstrate that LLMs can indeed generate content highly overlapping with copyrighted materials based on these partial inputs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_13831 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Measuring Copyright Risks of Large Language Model via Partial Information Probing Zhao, Weijie Shao, Huajie Xu, Zhaozhuo Duan, Suzhen Zhang, Denghui Computation and Language Artificial Intelligence Cryptography and Security Exploring the data sources used to train Large Language Models (LLMs) is a crucial direction in investigating potential copyright infringement by these models. While this approach can identify the possible use of copyrighted materials in training data, it does not directly measure infringing risks. Recent research has shifted towards testing whether LLMs can directly output copyrighted content. Addressing this direction, we investigate and assess LLMs' capacity to generate infringing content by providing them with partial information from copyrighted materials, and try to use iterative prompting to get LLMs to generate more infringing content. Specifically, we input a portion of a copyrighted text into LLMs, prompt them to complete it, and then analyze the overlap between the generated content and the original copyrighted material. Our findings demonstrate that LLMs can indeed generate content highly overlapping with copyrighted materials based on these partial inputs. |
| title | Measuring Copyright Risks of Large Language Model via Partial Information Probing |
| topic | Computation and Language Artificial Intelligence Cryptography and Security |
| url | https://arxiv.org/abs/2409.13831 |