AI News AI资讯 3h ago Updated 1h ago 更新于 1小时前 59

Anthropic’s $1.5B copyright settlement approved; only 350 authors opted out Anthropic 15亿美元版权和解案获批;仅350名作者退出

A US judge approved a $1.5 billion settlement between Anthropic and authors, marking the largest copyright class-action settlement ever reached. The settlement resolves claims regarding Anthropic's use of copyrighted books for AI training, with the court previously ruling such training constituted fair use but acknowledging potential issues with piracy. While the total settlement amount was upheld, the judge significantly reduced legal fees to under 7% ($101 million) and lowered service awards f Anthropic与作者达成15亿美元和解,终结了美国历史上认证的最大版权集体诉讼案。 法官批准和解协议,认定Anthropic使用书籍训练AI属于“合理使用”,但盗版行为可能侵权。 律师费从最初要求的20%大幅削减至低于7%(约1.01亿美元),原告代表服务费亦被调低。 超过91%的受影响作者和出版商已提交索赔申请,仅350人成功退出,绝大多数异议被驳回。

85
Hot 热度
75
Quality 质量
90
Impact 影响力

Analysis 深度分析

TL;DR

  • A US judge approved a $1.5 billion settlement between Anthropic and authors, marking the largest copyright class-action settlement ever reached.
  • The settlement resolves claims regarding Anthropic's use of copyrighted books for AI training, with the court previously ruling such training constituted fair use but acknowledging potential issues with piracy.
  • While the total settlement amount was upheld, the judge significantly reduced legal fees to under 7% ($101 million) and lowered service awards for lead plaintiffs, citing reasonableness standards.
  • Over 91% of affected authors and publishers have filed claims, and the court denied most late opt-out requests, emphasizing that the majority of the class supports the agreement.

Why It Matters

This settlement sets a critical precedent for the intersection of artificial intelligence development and intellectual property rights, establishing a financial framework for resolving copyright disputes without halting AI innovation. For AI practitioners and legal experts, it highlights the importance of fair use defenses while demonstrating that companies may still face significant liability for unauthorized data scraping or piracy. The outcome signals a shift toward negotiated resolutions in high-stakes IP litigation, offering a template for how the tech industry might handle similar conflicts with creative industries in the future.

Technical Details

  • Settlement Value: $1.5 billion total, with individual payouts estimated at approximately $3,000 per work, which the court noted is four times the minimum statutory damages.
  • Legal Fees Adjustment: Attorneys initially requested 20% ($300 million), later reduced to 12.5% ($187 million), but the judge capped fees at less than 7% ($101 million) to protect class interests.
  • Class Participation: Approximately 95% of the class received notifications, and 91% of impacted authors and publishers filed claims; only 350 members opted out.
  • Judicial Findings: US District Judge Araceli Martínez-Olguín ruled that Anthropic’s training on books was fair use but that its actions likely involved piracy, necessitating the settlement.
  • Post-Distribution Accounting: Lawyers must file a post-distribution accounting to ensure fees align with actual work performed, allowing for further reduction if necessary.

Industry Insight

  • Risk Mitigation Strategy: AI companies should prioritize robust licensing agreements and transparent data sourcing practices to avoid costly class-action lawsuits, even when fair use defenses exist.
  • Litigation Cost Management: The significant reduction in legal fees underscores the need for efficient litigation strategies and realistic budgeting for potential IP disputes, as courts are increasingly scrutinizing attorney compensation in large settlements.
  • Creator Relations: The overwhelming participation in the settlement suggests that many creators prefer guaranteed compensation over uncertain litigation outcomes, indicating a potential market for standardized licensing models between AI firms and content creators.

TL;DR

  • Anthropic与作者达成15亿美元和解,终结了美国历史上认证的最大版权集体诉讼案。
  • 法官批准和解协议,认定Anthropic使用书籍训练AI属于“合理使用”,但盗版行为可能侵权。
  • 律师费从最初要求的20%大幅削减至低于7%(约1.01亿美元),原告代表服务费亦被调低。
  • 超过91%的受影响作者和出版商已提交索赔申请,仅350人成功退出,绝大多数异议被驳回。

为什么值得看

该案件确立了AI公司利用受版权保护内容进行训练的法律边界,对生成式AI行业的合规性具有里程碑意义。高额和解金及律师费的调整反映了司法系统对AI侵权赔偿规模的重新评估,为后续类似诉讼提供了重要参考。

技术解析

  • 法律定性:法院裁定Anthropic将书籍用于AI模型训练构成“合理使用”(Fair Use),但明确区分了训练过程与潜在的盗版分发行为,后者可能构成侵权。
  • 赔偿机制:和解基金总额15亿美元,预计每部作品赔付约3,000美元,这一金额是法定最低损害赔偿的四倍。若最终有效索赔总额低于基金规模,剩余资金将重新分配。
  • 程序细节:法官Araceli Martínez-Olguín驳回了大多数异议和逾期退出请求,仅因“可原谅的疏忽”(如未收到通知、重病、语言障碍)批准了极少数例外情况。
  • 费用结构:律师费经法院严格审查后,从原计划的3亿美元降至1.01亿美元以下,并要求在资金分配完成后进行事后审计,以防虚报工时。

行业启示

  • 合规成本显性化:AI巨头需将版权清算纳入核心运营成本,建立更透明的内容来源追踪机制,以应对日益严格的知识产权监管。
  • “合理使用”并非绝对护身符:尽管训练阶段被认定为合理使用,但企业仍需警惕数据分发环节的法律风险,避免被视为盗版渠道。
  • 诉讼策略转向和解:面对庞大的潜在索赔群体,通过大规模和解快速结案成为科技公司的首选策略,以减少长期法律不确定性并重塑公众形象。

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Claude Claude Legal AI 法律AI Policy 政策 Regulation 监管 Ethics 伦理