| 人形机器人训练数据合理使用的规范架构与制度展开 |
 点此下载全文 |
| 引用本文：常柏1，全荃2.人形机器人训练数据合理使用的规范架构与制度展开[J].财经理论与实践,2026,(1):152-160 |
| 摘要点击次数: 7 |
| 全文下载次数: 4 |
|
|
| 中文摘要:人形机器人作为具身智能的典型形态，兼具认知感知、情绪交互与动作执行能力，其训练过程对多模态、高语义数据的依赖程度显著提升，尤其是对人体动作存在直接模仿性，形成区别于一般人工智能的数据使用模式。此种数据训练的目的和结果与原始数据存在完全不同的消费卖点，同时面向不同的消费者群体，对原始数据商品的市场破坏性有限。但现行著作权法合理使用条款对“附随性复制”“转换性使用”与“非替代性市场影响”等新型使用情形缺乏明确规范，难以有效回应人形机器人训练所引发的权利边界问题。人形机器人训练数据的表达解构机制及功能转换逻辑，可在著作权法现有体系中识别出新的合理使用空间。制度设计上，短期应借助实施条例的扩张性规范路径，结合行政监管、行业机制，实现法律适应性的有序进化；长期可通过条文修订扩大合理使用类型，并引入开发者的信息披露义务与合规保障机制，以平衡技术创新、著作权保护之间的矛盾。 |
| 中文关键词:人形机器人 具身智能 训练数据 著作权侵权 合理使用 |
| |
| The Normative Framework and System for the Rational Use of Training Data for Humanoid Robots |
|
|
| Abstract:As a quintessential form of embodied intelligence, humanoid robots possess cognitive perception, emotional interaction, and action execution capabilities. Their training process significantly increases reliance on multimodal, high-semantic data, particularly exhibiting direct imitation of human movements, thereby forming a distinct data usage pattern compared to general artificial intelligence. The purpose and outcomes of this data training present entirely distinct consumption value propositions compared to the original data, targeting different consumer groups. Consequently, its disruptive impact on the original data commodity market is limited. However, current copyright law’s fair use provisions lack clear regulations for novel usage scenarios such as “incidental reproduction”“transformative use” and “non-substitutive market impact” making it difficult to effectively address the boundary issues raised by humanoid robot training. By examining the expressive deconstruction mechanisms and functional conversion logic of humanoid robot training data, new fair use spaces can be identified within the existing copyright framework. Institutionally, short-term solutions should leverage expansive regulatory pathways through implementing regulations, combined with administrative oversight and industry mechanisms, to achieve orderly adaptive evolution of the law. Long-term approaches should involve amending statutory provisions to broaden fair use categories while introducing developers’ information disclosure obligations and compliance safeguards to balance technological innovation with copyright protection. |
| keywords:humanoid robot embodied intelligence training data copyright infringement fair use |
| 查看全文 查看/发表评论 下载pdf阅读器 |
|
|
|
位访问者,今日访问
