TY - GEN
T1 - CoProtector
T2 - 31st ACM Web Conference, WWW 2022
AU - Sun, Zhensu
AU - Du, Xiaoning
AU - Song, Fu
AU - Ni, Mingze
AU - Li, Li
N1 - Publisher Copyright:
© 2022 ACM.
PY - 2022/4/25
Y1 - 2022/4/25
N2 - Github Copilot, trained on billions of lines of public code, has recently become the buzzword in the computer science research and practice community. Although it is designed to help developers implement safe and effective code with powerful intelligence, practitioners and researchers raise concerns about its ethical and security problems, e.g., should the copyleft licensed code be freely leveraged or insecure code be considered for training in the first place? These problems pose a significant impact on Copilot and other similar products that aim to learn knowledge from large-scale open-source code through deep learning models, which are inevitably on the rise with the fast development of artificial intelligence. To mitigate such impacts, we argue that there is a need to invent effective mechanisms for protecting open-source code from being exploited by deep learning models. Here, we design and implement a prototype, CoProtector, which utilizes data poisoning techniques to arm source code repositories for defending against such exploits. Our large-scale experiments empirically show that CoProtector is effective in achieving its purpose, significantly reducing the performance of Copilot-like deep learning models while being able to stably reveal the secretly embedded watermark backdoors.
AB - Github Copilot, trained on billions of lines of public code, has recently become the buzzword in the computer science research and practice community. Although it is designed to help developers implement safe and effective code with powerful intelligence, practitioners and researchers raise concerns about its ethical and security problems, e.g., should the copyleft licensed code be freely leveraged or insecure code be considered for training in the first place? These problems pose a significant impact on Copilot and other similar products that aim to learn knowledge from large-scale open-source code through deep learning models, which are inevitably on the rise with the fast development of artificial intelligence. To mitigate such impacts, we argue that there is a need to invent effective mechanisms for protecting open-source code from being exploited by deep learning models. Here, we design and implement a prototype, CoProtector, which utilizes data poisoning techniques to arm source code repositories for defending against such exploits. Our large-scale experiments empirically show that CoProtector is effective in achieving its purpose, significantly reducing the performance of Copilot-like deep learning models while being able to stably reveal the secretly embedded watermark backdoors.
KW - data poisoning
KW - dataset protection
KW - deep learning
KW - open-source code
UR - https://www.scopus.com/pages/publications/85129794072
U2 - 10.1145/3485447.3512225
DO - 10.1145/3485447.3512225
M3 - 会议稿件
AN - SCOPUS:85129794072
T3 - WWW 2022 - Proceedings of the ACM Web Conference 2022
SP - 652
EP - 660
BT - WWW 2022 - Proceedings of the ACM Web Conference 2022
PB - Association for Computing Machinery, Inc
Y2 - 25 April 2022 through 29 April 2022
ER -