社交媒体巨头Meta Platforms Inc. 在其人工智能研发过程中，涉嫌未经授权使用“盗版”书籍来训练其人工智能模型。这一行为已经引起了版权方面的争议。
根据法庭文件，Meta在其人工智能项目Llama 1和Llama 2的研发中使用了名为Books3的数据集。Books3是一个包含了近20万本书的纯文本集合，总容量近37GB，是一个知名的开源图书数据集。然而，Books3中包含了大量受版权保护的作品。
Meta在一份声明中承认使用了Books3数据集，但辩称其行为属于“合理使用”，因此无需为此付费。Meta主张，其使用受版权保护的作品来训练大模型不需要“同意、许可或付费”。
这一立场在版权法方面引发了争议。一般来说，使用受版权保护的作品进行复制、分发或改编，都需要获得版权持有人的许可并支付相应的费用。然而，Meta认为，其使用受版权保护的作品是为了教育和研究目的，属于“合理使用”的范畴。
这起诉讼案将人工智能领域的版权问题推向了风口浪尖。随着人工智能技术的不断发展，越来越多的公司和研究机构开始使用大量的数据集来训练其人工智能模型。然而，如何确保这些数据集中的内容不侵犯版权，成为了一个亟待解决的问题。
英文标题：Meta’s AI Accused of Infringing Copyrights by Using “Pirated” Books
Keywords: Meta, Artificial Intelligence, Copyright Infringement
News content:
Social media behemoth Meta Platforms Inc. is涉嫌using “pirated” books to train its artificial intelligence (AI) models, sparking a copyright controversy.
According to court documents, Meta used a dataset called Books3 in the development of its AI projects Llama 1 and Llama 2. Books3 is a collection of nearly 200,000 books in text form, with a total capacity of nearly 37GB, and is a well-known open-source book dataset. However, it contains numerous copyright-protected works.
Meta admitted using the Books3 dataset in a statement but claimed that its actions constitute “fair use,” Therefore, it argued, it does not need “consent, licensing, or payment” for using copyrighted works to train large models.
This position has sparked controversy under copyright law. Generally, using copyrighted works for copying, distribution, or adaptation requires obtaining the authorization of the copyright holder and paying the corresponding fees. However, Meta contends that its use of copyrighted works for educational and research purposes is within the scope of “fair use.”
This copyright lawsuit pushes the boundaries of AI ethics. As AI technology continues to advance, more and more companies and research institutions are using large amounts of datasets to train their AI models. However, ensuring that the content within these datasets does not infringe on copyrights has become an issue that needs to be addressed urgently.
【来源】https://www.techspot.com/news/101507-meta-admits-using-pirated-books-train-ai-but.html
Views: 1