**Meta公司被曝使用“盗版”书籍训练AI，引发版权争议**
近日，全球知名科技公司Meta在一桩诉讼中承认，其用于训练人工智能（AI）的大量数据来源于一个名为“Books3”的数据集。这一消息引发了关于版权和“合理使用”原则的广泛讨论。
据了解，“Books3”是一个知名的开源图书数据集，包含近20万本书的纯文本集合，总容量接近37GB。但Meta在诉讼中辩称，尽管这些书籍受版权保护，但其使用这些作品来训练大模型并不需要得到作者或版权持有者的“同意、许可或付费”。
Meta进一步主张，任何未经授权复制“Books3”中受版权保护的作品都应被视为“合理使用”。这一立场引起了法律界和公众的广泛关注。一方面，有人认为Meta的观点为AI研究提供了新的可能性，另一方面，也有人担忧这可能会为未来的版权侵权行为开辟道路。
TechSpot报道指出，随着AI技术的日益普及，如何平衡创新与版权保护之间的关系成为了一个亟待解决的问题。此次事件无疑为这一问题提供了一个重要的案例，未来可能会对整个AI产业产生深远的影响。
英语如下：
====
“News Headline: Meta Trains AI Using “Pirated” Books, Refus====
“News Headline: Meta Trains AI Using “Pirated” Books, Refusal to Pay Raises Controversy
Keywords: Meta, pirated books, fair use
News Content: **Meta Company Accused of Using “Pirated” Books to Train AI, Causing Copyright Controversy**
Recently, global technology giant Meta admitted in a lawsuit that a large amount of data used to train its Artificial Intelligence (AI) came from a dataset called “Books3.” This news has sparked widespread discussion about copyright and the principle of “fair use”.
It is understood that “Books3” is a well-known open-source book dataset, containing nearly 200,000 books’ plain text collection with a total capacity of nearly 37GB. However, Meta argued in the lawsuit that although these books are protected by copyright, their use to train large models does not require the “consent, permission, or payment” of the authors or copyright holders.
Meta further contends that any unauthorized reproduction of copyrighted works in “Books3” should be considered “fair use”. This position has attracted wide attention from the legal community and the public. On one hand, some believe that Meta’s viewpoint provides new possibilities for AI research. On the other hand, there are concerns that this may pave the way for future copyright infringements.
TechSpot reports that as AI technology becomes increasingly popular, how to balance innovation and copyright protection has become an urgent issue to be solved. This incident undoubtedly provides an important case study for this issue and may have far-reaching implications for the entire AI industry in the future.”
【来源】https://www.techspot.com/news/101507-meta-admits-using-pirated-books-train-ai-but.html
Views: 1