As the use of artificial intelligence technology, particularly deep learning models like the diffusion model, becomes more prevalent in the creation of art, concerns regarding the use of unauthorized work samples in their training have emerged. The lack of supervision and relevant laws has contributed to the problem. This study examines the potential infringement issues that may arise during the training of the diffusion model and explores the legality of using unauthorized samples for deep learning model training. While some scholars argue that copyright law only protects expression and not painting style, therefore, using unauthorized works in model training is not considered infringement, we propose a different viewpoint. By considering the essence of artificial intelligence from an information theory perspective, we highlight that it is still a deterministic algorithm and that data processing does not bring about an increase in information entropy without the input of additional information. Thus, the painting created by the diffusion model is essentially a mash-up of paintings in its training space, and as such, it is a copy or adaptation of the original work and should be licensed by the creator. We highlight the critical difference between human learning and AI "learning," emphasizing the need for effective protection and encouragement of human artistic innovation while embracing the wave of AI.

Shumin Wang. Analyzing Copyright Infringement by Artificial Intelligence: The Case of the Diffusion Model. Academic Journal of Humanities & Social Sciences (2023) Vol. 6, Issue 6: 1-3. https://doi.org/10.25236/AJHSS.2023.060601.

[1] van Dis E. A., Bollen J., Zuidema W., van Rooij R., & Bockting C. L. (2023). ChatGPT: five priorities for research. Nature, 614(7947), 224-226.

[2] Li Q., Cai W., Wang X., Zhou Y., Feng D. D., & Chen M. (2014, December). Medical image classification with convolutional neural network. In 2014 13th international conference on control automation robotics & vision (ICARCV) (pp. 844-848). IEEE.

[3] Solum L. B. (1991). Legal personhood for artificial intelligences. NCL Rev., 70, 1231.

[4] Naqvi Z. (2020). Artificial intelligence, copyright, and copyright infringement. Marq. Intell. Prop. L. Rev., 24, 15.

[5] Carlini N., Hayes J., Nasr M., Jagielski M., Sehwag V., Tramèr F., & Wallace E. (2023). Extracting training data from diffusion models. arXiv preprint arXiv:2301.13188.

[6] Thomas D. R., Pastrana S., Hutchings A., Clayton R., & Beresford A. R. (2017, November). Ethical issues in research using datasets of illicit origin. In Proceedings of the 2017 Internet Measurement Conference (pp. 445-462).

[7] Smith L. N., & Chambers R. Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License.

[8] Hornbeck P. V., Chabra I., Kornhauser J. M., Skrzypek E., & Zhang B. (2004). PhosphoSite: A bioinformatics resource dedicated to physiological protein phosphorylation. Proteomics, 4(6), 1551- 1561.

[9] Krizhevsky A., & Hinton G. (2009). Learning multiple layers of features from tiny images.

[10] Schlachter E. (1997). The intellectual property renaissance in cyberspace: why copyright law could be unimportant on the Internet. Berkeley Tech. LJ, 12, 15.

[11] Ruipérez C., Gutiérrez E., Puente C., & Olivas J. A. (2017). New Challenges of Copyright Authorship in AI. In Proceedings on the International Conference on Artificial Intelligence (ICAI) (pp. 291-296). The Steering Committee of the World Congress in Computer Science, Computer Engineering and Applied Computing (WorldComp).

[12] Lazar J. C. (1995). Protecting ideas and ideals: Copyright law in the People's Republic of China. Law & Pol'y Int'l Bus., 27, 1185.

[13] Jones,R. H. (1990). The myth of the idea/expression dichotomy in copyright law. Pace L. Rev., 10, 551.

[14] Kullback S. (1997). Information theory and statistics. Courier Corporation.

[15] Gillotte J. L. (2019). Copyright infringement in ai-generated artworks. UC Davis L. Rev., 53, 2655