Learning to Poison Large Language Models for Downstream Manipulation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916765675552768 |
|---|---|
| author | Zhou, Xiangyu Qiang, Yao Zade, Saleh Zare Roshani, Mohammad Amin Khanduri, Prashant Zytko, Douglas Zhu, Dongxiao |
| author_facet | Zhou, Xiangyu Qiang, Yao Zade, Saleh Zare Roshani, Mohammad Amin Khanduri, Prashant Zytko, Douglas Zhu, Dongxiao |
| contents | The advent of Large Language Models (LLMs) has marked significant achievements in language processing and reasoning capabilities. Despite their advancements, LLMs face vulnerabilities to data poisoning attacks, where the adversary inserts backdoor triggers into training data to manipulate outputs. This work further identifies additional security risks in LLMs by designing a new data poisoning attack tailored to exploit the supervised fine-tuning (SFT) process. We propose a novel gradient-guided backdoor trigger learning (GBTL) algorithm to identify adversarial triggers efficiently, ensuring an evasion of detection by conventional defenses while maintaining content integrity. Through experimental validation across various language model tasks, including sentiment analysis, domain generation, and question answering, our poisoning strategy demonstrates a high success rate in compromising various LLMs' outputs. We further propose two defense strategies against data poisoning attacks, including in-context learning (ICL) and continuous learning (CL), which effectively rectify the behavior of LLMs and significantly reduce the decline in performance. Our work highlights the significant security risks present during SFT of LLMs and the necessity of safeguarding LLMs against data poisoning attacks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_13459 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning to Poison Large Language Models for Downstream Manipulation Zhou, Xiangyu Qiang, Yao Zade, Saleh Zare Roshani, Mohammad Amin Khanduri, Prashant Zytko, Douglas Zhu, Dongxiao Machine Learning Computation and Language Cryptography and Security The advent of Large Language Models (LLMs) has marked significant achievements in language processing and reasoning capabilities. Despite their advancements, LLMs face vulnerabilities to data poisoning attacks, where the adversary inserts backdoor triggers into training data to manipulate outputs. This work further identifies additional security risks in LLMs by designing a new data poisoning attack tailored to exploit the supervised fine-tuning (SFT) process. We propose a novel gradient-guided backdoor trigger learning (GBTL) algorithm to identify adversarial triggers efficiently, ensuring an evasion of detection by conventional defenses while maintaining content integrity. Through experimental validation across various language model tasks, including sentiment analysis, domain generation, and question answering, our poisoning strategy demonstrates a high success rate in compromising various LLMs' outputs. We further propose two defense strategies against data poisoning attacks, including in-context learning (ICL) and continuous learning (CL), which effectively rectify the behavior of LLMs and significantly reduce the decline in performance. Our work highlights the significant security risks present during SFT of LLMs and the necessity of safeguarding LLMs against data poisoning attacks. |
| title | Learning to Poison Large Language Models for Downstream Manipulation |
| topic | Machine Learning Computation and Language Cryptography and Security |
| url | https://arxiv.org/abs/2402.13459 |