内容来源：https://www.theverge.com/news/775230/anthropic-piracy-class-action-lawsuit-settlement-rejected

内容总结：

美国人工智能公司Anthropic拟以15亿美元和解图书侵权集体诉讼一案出现变数。旧金山联邦法院法官威廉·阿尔苏普本周叫停了这项和解协议，理由是担心原告律师"在闭门谈判中将协议强加给作家群体"。

这起集体诉讼指控Anthropic未经授权使用数十万本受版权保护的书籍训练其AI模型。根据和解条款，每位作者每部被侵权的作品可获得约3000美元赔偿，预计涉及46.5万部作品。

阿尔苏普法官虽曾裁定Anthropic使用合法购买书籍训练AI属于合理使用，但强调使用非法下载内容仍需承担责任。他表示需要更详细审查索赔流程，并要求提供确切的涉事作品数量，以避免潜在后续诉讼。法官要求必须确保所有集体诉讼成员获得充分告知。

美国出版商协会首席执行官玛丽亚·帕兰特批评法官"未能理解出版行业运作机制"，强调集体诉讼本应解决争端而非制造新矛盾。原告律师则承诺将确保每位合法索赔者获得补偿。

此案将于9月25日再次开庭审理。阿尔苏普法官表示将重新评估是否批准该协议。

中文翻译：

美国联邦法院法官近日叫停了人工智能公司Anthropic涉及图书侵权案的15亿美元和解协议，因主审这起集体诉讼案的法官对和解条款提出质疑。据彭博法律和美联社报道，主审法官威廉·阿尔苏普在本周的听证会上驳回了该和解协议，担心集体诉讼律师会"强行迫使作者接受"闭门达成的交易。

这起集体诉讼源于美国作家指控Anthropic使用数十万本受版权保护的书籍训练AI模型。虽然阿尔苏普法官此前裁定Anthropic使用合法购买书籍训练AI属于合理使用，但使用非法下载作品仍需承担责任，因此允许诉讼继续进行。

除担心作者被迫接受协议外，阿尔苏普法官表示需要审查和解协议中索赔流程的详细信息。他坦言："面对这么巨额的资金，我对那些伺机分一杯羹的人感到不安。"根据协议，每位作者每部被侵权的作品可获得约3000美元赔偿。原告律师称涉及作品约46.5万部，但法官要求提供确切数字以避免潜在纠纷。

美国出版商协会首席执行官玛丽亚·帕兰特批评法官"缺乏对出版业运作方式的了解"，强调集体诉讼本应解决争端而非制造新矛盾。原告律师贾斯汀·尼尔森则承诺将确保所有合法索赔获得补偿。

阿尔苏普法官将于9月25日再次举行听证会审议此事，并表示将最终决定是否"勉强批准该协议"。

英文来源：

Anthropic’s $1.5 billion book piracy settlement has been put on pause after the federal judge overseeing the class action case raised concerns about the terms of the agreement. During a hearing this week, Judge William Alsup rejected the settlement over concerns that class action lawyers will create a deal behind closed doors that they will force “down the throats of authors,” according to reports from Bloomberg Law and the Associated Press.
Judge puts Anthropic’s $1.5 billion book piracy settlement on hold
A federal judge has raised concerns about the terms of the settlement, which would give authors around $3,000 for each book Anthropic is accused of stealing.
A federal judge has raised concerns about the terms of the settlement, which would give authors around $3,000 for each book Anthropic is accused of stealing.
Anthropic agreed to pay the landmark settlement last week, putting to rest a class action lawsuit from US authors that accused the AI company of training its models on hundreds of thousands of copyrighted books. Judge Alsup let the class action lawsuit move forward after ruling that Anthropic training its AI models on purchased books counts as fair use, but that it could be liable for training on illegally downloaded work.
In addition to his concerns over authors being strong-armed into signing a deal, Alsup says he needs to review more information about the claims process outlined in the settlement. “I have an uneasy feeling about hangers on with all this money on the table,” Alsup said, as reported by Bloomberg Law.
Under the settlement, authors and publishers would receive about $3,000 for covered works. As noted by AP, an attorney for the authors said there are around 465,000 books that would be covered by the settlement, but Judge Alsup asked for a solid number to ensure Anthropic doesn’t get hit with other lawsuits “coming out of the woodwork.” He added that class members will need to be given “very good notice” to make sure they’re aware of the case.
Maria Pallante, CEO of the Association of American Publishers, an industry group backing the authors’ lawsuit, told AP that Alsup “demonstrated a lack of understanding of how the publishing industry works.” Pallante said that “class actions are supposed to resolve cases, not create new disputes, and certainly not between the class members who were harmed in the first place.” The authors’ attorney, Justin Nelson, said in a statement to Bloomberg Law that the lawyers “care deeply that every single proper claim gets compensation.”
Judge Alsup will revisit the settlement during another hearing on September 25th. “We’ll see if I can hold my nose and approve it,” Alsup said, according to the AP.