本文主要是介绍【Datawhale 大模型基础】第五章 大模型法律，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
第五章 大模型法律
When training LLMs, we must address issues of copyright and fair use. Obtaining permissions from every individual due to the indiscriminate nature of web scraping would be extremely challenging. The generative nature of the model may pose challenges to the debate on fair use (it can compete with humans).
And for this topic, I want to find an interesting but vital question: should AI-generated paintings be recognized as works and be protected by copyright law?
Recently, the Beijing Internet Court issued a first-instance judgment on a copyright infringement dispute related to AI-generated images (AI paintings). It is reported that this case is the first case in the field of AI-generated images related to copyright.
The plaintiff, Mr. Li, used AI to generate the images in question and posted them on the Xiaohongshu platform. The defendant is a Baijiahao blogger who used the plaintiff’s AI-generated images as illustrations for an article, leading to the plaintiff’s lawsuit.
The Beijing Internet Court held that the AI-generated images in question possess the “originality” requirement, reflecting the creative intellectual input of a human and should be recognized as works protected by copyright law.
- Regarding the determination of intellectual achievements: "From the plaintiff’s conception of the images in question to the final selection of the images, the plaintiff has made a certain intellectual input throughout the entire process, such as designing the presentation of characters, selecting prompts, arranging the order of prompts, setting related parameters, and selecting which image meets the expectations, etc. The images in question reflect the plaintiff’s intellectual input, thus meeting the requirement of “intellectual achievements.”
- Regarding the determination of “originality”: "The plaintiff designed the visual elements such as characters and their presentation through prompts, and set the layout and composition of the images through parameters, demonstrating the plaintiff’s choices and arrangements. On the other hand, after obtaining the first image by inputting prompts and setting related parameters, the plaintiff continued to add prompts, modify parameters, continuously adjust and revise, and finally obtained the images in question. This process of adjustment and revision also reflects the plaintiff’s aesthetic choices and personal judgment… The images in question are not a “mechanical intellectual achievement.” In the absence of contrary evidence, it can be determined that the plaintiff independently completed the images in question, demonstrating the plaintiff’s personalized expression. Therefore, the images in question meet the requirement of “originality.”
- Regarding the determination of the work: “When people use artificial intelligence models to generate images… fundamentally, it is still people using tools to create… that is, the intellectual input throughout the entire creative process comes from humans rather than artificial intelligence models. Encouraging creation is recognized as the core purpose of the copyright system… Artificial intelligence-generated images, as long as they can reflect the creative intellectual input of humans, should be recognized as works and protected by copyright law.”
- Regarding the determination of artistic works: "The images in question are two-dimensional artistic works composed of lines and colors with aesthetic significance, belonging to works of fine art. At the same time, when the images in question can be attributed to specific types of works, there is no need to apply the “other works provision” for protection, as they do not belong to “other intellectual achievements that meet the characteristics of works.”
- Regarding the determination of copyright: “The plaintiff directly made relevant settings to the AI model in question according to the needs and ultimately selected the images in question. The images in question were directly produced based on the plaintiff’s intellectual input and reflect the plaintiff’s personalized expression. Therefore, the plaintiff is the author of the images in question and enjoys the copyright to the images in question.”
For further study about it, can see the original judgment from: https://mp.weixin.qq.com/s/Wu3-GuFvMJvJKJobqqq7vQ, it is a landmark decisions, and I strongly recommend that every reader read the original judgment.
END
这篇关于【Datawhale 大模型基础】第五章 大模型法律的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
