论人工智能提示词的著作权法保护
Copyright Protection of Artificial Intelligence Prompts
摘要: 生成式人工智能的发展使提示词成为人类介入人工智能生成过程的重要媒介。随着提示词由简单关键词、功能指令逐渐发展为包含场景设计、语言风格、结构安排和审美要求的复杂文本,提示词本身是否属于著作权法意义上的作品,成为新的法律问题。现有研究较多从提示词对人工智能生成内容的控制程度出发,讨论提示词对生成内容可版权性的影响,但提示词本身的著作权属性与基于提示词生成内容的著作权问题并不相同。提示词能够影响生成结果,并不意味着其自身当然具有独创性。生成内容是否构成作品,同样也不能决定提示词的法律属性。本文以提示词本身为研究对象,在思想与表达二分原则以及独创性标准的基础上,分析提示词由“指令”向“表达”转化的条件。本文认为,提示词的可版权性应以具体表达是否具有独创性为核心。对于简单指令、一般创作构思和功能性参数,应排除著作权保护;对于具有较高选择、组合、编排和个性化表达程度的复杂提示词,则可以根据其具体表现形式认定为作品,并给予范围有限的著作权保护。同时,应避免通过提示词著作权间接垄断创作理念、生成方法和技术规则,并在必要情况下以商业秘密和反不正当竞争法提供补充保护。
Abstract: Generative artificial intelligence has made prompts an important means of human intervention in the generation of content. As prompts have developed from simple keywords and functional instructions into complex texts containing scene design, linguistic style, structural arrangement, and aesthetic requirements, a new copyright question has emerged: can a prompt itself qualify as a work under copyright law? Much of the existing literature examines prompts in relation to the copyrightability of AI-generated content, particularly the extent to which a prompt determines or controls the resulting output. The copyright status of the prompt itself, however, is a separate question. A prompt’s influence on the generated output does not necessarily establish originality in the prompt, nor does the copyright status of the output determine the legal status of the prompt. Against this background, this article considers when a prompt may move beyond an “instruction” and constitute protectable “expression,” with particular reference to the idea-expression dichotomy and the originality requirement. The copyrightability of a prompt should turn on the originality of its specific expression. Simple instructions, general creative ideas, and functional parameters should remain outside the scope of copyright protection. More complex prompts, where the author’s choices, combinations, arrangements, and individualized expression are sufficiently apparent, may qualify as works depending on their particular form and receive copyright protection within an appropriate scope. Such protection should not, however, extend to creative ideas, methods of generation, or technical rules merely because they are embodied in a prompt. Where appropriate, trade secret protection and the law against unfair competition may serve as supplementary means of protection.
参考文献
|
[1]
|
杜维娜. AI提示词保护的理论构想与路径选择[J]. 河北法学, 2026, 44(6): 128-144.
|
|
[2]
|
丛立先, 徐子文. 生成式人工智能提示词的可版权性辨析[J]. 知识产权, 2026(4): 60-72.
|
|
[3]
|
易玲, 王静. 论人工智能生成内容著作权法保护[J]. 湘潭大学学报(哲学社会科学版), 2019, 43(6): 69-73.
|
|
[4]
|
王迁. 人工智能生成内容与著作权保护[J]. 版权理论与实务, 2023(7): 3-18.
|
|
[5]
|
管依萌. AI提示词的版权之争[N]. 法治周末, 2026-05-28(005).
|
|
[6]
|
毕文轩. 生成式人工智能生成内容的版权属性与保护路径[J]. 比较法研究, 2024(3): 55-71.
|
|
[7]
|
陈亮, 朱晟男. 激励前置与梯度赋权: 生成式人工智能提示词保护的初步设想[J]. 北京工商大学学报(社会科学版), 2025, 40(4): 131-142.
|
|
[8]
|
王迁. 人工智能生成的内容是作品吗?——以学术规范和著作权法的关系为视角[J]. 中国法律评论, 2025(5): 36-51.
|
|
[9]
|
杨力, 梁子卿. 人工智能作品是否应受著作权保护——从提示词干预到比较法路径的制度反思[J]. 北京政法职业学院学报, 2025(2): 57-64.
|