联邦法院于2026年5月裁定，Anthropic非法从影子图书馆抓取了10多位斯坦福大学教授的受版权保护书籍，用于训练其Claude AI模型。
Stanford Professors Challenge Anthropic Over Book Piracy in Claude AI Training A federal court ruled in May 2026 that Anthropic illegally scraped copyrighted books by over 10 Stanford professors from shadow libraries to train its Claude AI models.
在Bartz诉Anthropic案达成的15亿美元和解中，William Alsup法官禁止Anthropic通过盗版作品来维护核心图书馆，同时将合法获取的训练数据归类为合理使用。
In the resulting $1.5 billion settlement in Bartz v. Anthropic, Judge William Alsup barred Anthropic from pirating works to maintain a central library, while classifying legally acquired training data under fair use.
作者Tobias Wolff和Chang-Rae Lee选择退出和解，并提起单独诉讼Cruz诉Anthropic案，寻求每部被侵权作品高达15万美元的法定赔偿金以及永久禁令。这场纠纷反映了围绕知识产权与AI发展的更广泛法律张力。
Authors Tobias Wolff and Chang-Rae Lee opted out of the settlement and filed a separate lawsuit, Cruz v. Anthropic, seeking statutory damages up to $150,000 per infringed work alongside a permanent injunction. The dispute reflects broader legal tensions surrounding intellectual property and AI development.