Disguised Copyright Infringement of Latent Diffusion Models

Yiwei Lu, Matthew Y. R. Yang, Zuoqiu Liu, Gautam Kamath, Yaoliang Yu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:33196-33218, 2024.

Abstract

Copyright infringement may occur when a generative model produces samples substantially similar to some copyrighted data that it had access to during the training phase. The notion of access usually refers to including copyrighted samples directly in the training dataset, which one may inspect to identify an infringement. We argue that such visual auditing largely overlooks a concealed copyright infringement, where one constructs a disguise that looks drastically different from the copyrighted sample yet still induces the effect of training Latent Diffusion Models on it. Such disguises only require indirect access to the copyrighted material and cannot be visually distinguished, thus easily circumventing the current auditing tools. In this paper, we provide a better understanding of such disguised copyright infringement by uncovering the disguises generation algorithm, the revelation of the disguises, and importantly, how to detect them to augment the existing toolbox. Additionally, we introduce a broader notion of acknowledgment for comprehending such indirect access.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-lu24m, title = {Disguised Copyright Infringement of Latent Diffusion Models}, author = {Lu, Yiwei and Yang, Matthew Y. R. and Liu, Zuoqiu and Kamath, Gautam and Yu, Yaoliang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {33196--33218}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/lu24m/lu24m.pdf}, url = {https://proceedings.mlr.press/v235/lu24m.html}, abstract = {Copyright infringement may occur when a generative model produces samples substantially similar to some copyrighted data that it had access to during the training phase. The notion of access usually refers to including copyrighted samples directly in the training dataset, which one may inspect to identify an infringement. We argue that such visual auditing largely overlooks a concealed copyright infringement, where one constructs a disguise that looks drastically different from the copyrighted sample yet still induces the effect of training Latent Diffusion Models on it. Such disguises only require indirect access to the copyrighted material and cannot be visually distinguished, thus easily circumventing the current auditing tools. In this paper, we provide a better understanding of such disguised copyright infringement by uncovering the disguises generation algorithm, the revelation of the disguises, and importantly, how to detect them to augment the existing toolbox. Additionally, we introduce a broader notion of acknowledgment for comprehending such indirect access.} }
Endnote
%0 Conference Paper %T Disguised Copyright Infringement of Latent Diffusion Models %A Yiwei Lu %A Matthew Y. R. Yang %A Zuoqiu Liu %A Gautam Kamath %A Yaoliang Yu %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-lu24m %I PMLR %P 33196--33218 %U https://proceedings.mlr.press/v235/lu24m.html %V 235 %X Copyright infringement may occur when a generative model produces samples substantially similar to some copyrighted data that it had access to during the training phase. The notion of access usually refers to including copyrighted samples directly in the training dataset, which one may inspect to identify an infringement. We argue that such visual auditing largely overlooks a concealed copyright infringement, where one constructs a disguise that looks drastically different from the copyrighted sample yet still induces the effect of training Latent Diffusion Models on it. Such disguises only require indirect access to the copyrighted material and cannot be visually distinguished, thus easily circumventing the current auditing tools. In this paper, we provide a better understanding of such disguised copyright infringement by uncovering the disguises generation algorithm, the revelation of the disguises, and importantly, how to detect them to augment the existing toolbox. Additionally, we introduce a broader notion of acknowledgment for comprehending such indirect access.
APA
Lu, Y., Yang, M.Y.R., Liu, Z., Kamath, G. & Yu, Y.. (2024). Disguised Copyright Infringement of Latent Diffusion Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:33196-33218 Available from https://proceedings.mlr.press/v235/lu24m.html.

Related Material