Anthropic secretly used 7 million books to train Claude, facing trillion-dollar piracy lawsuits! The mad dash of AI giants and the legal boundaries.

Anthropic is accused of using 7 million pirated books to train models, facing a collective lawsuit from authors that could result in over a trillion dollars in damages, impacting data compliance and investment evaluations across the entire AI industry. (Background: AI privacy collapse "ChatGPT conversation" exposed in front of the law; Altman: I fear inputting personal data, hard to know who will access the information) (Supplementary background: Can AI overturn cases? A lawyerless woman used ChatGPT to unearth $5 million in inheritance fraud, persuading the court to restart investigations. A copyright lawsuit concerning 7 million books is putting Anthropic, the parent company behind the popular AI model Claude, under the spotlight. In late July, author groups officially sued Anthropic for using a large number of copyrighted books to train its large language model. If defeated, Anthropic could face astronomical compensation. The judge ruled: training activities can be considered fair use, but data sources cannot. According to a report by Fortune, Federal Judge William Alsup determined that if AI companies "legally obtain" books and convert the content into parameters for model training, it could constitute fair use; however, if the data is illegally downloaded from shadow libraries like LibGen or PiLiMi, the situation reverses, and the training results cannot be exempted. This delineation means that while Anthropic gains some support in its model development methods, it must independently face scrutiny over its data sources. Compensation estimation: a survival test looking up to $1 trillion. U.S. copyright law stipulates that statutory damages for each infringing work range from $750 to $150,000; if the jury determines the actions were "willful," the maximum will apply. Legal scholar Ed Lee cited this standard to estimate that if 6 million books were infringed, Anthropic could face a maximum compensation of $1.05 trillion. Compared to the company's approximately $4 billion annual revenue and a valuation of $60 billion to $100 billion, this is not only an astronomical figure but also poses a direct threat to its operational continuity. Legal discrepancies and industry ripple effects. It is noteworthy that Judge Vince Chhabria, in another data lawsuit involving Meta in the Northern California federal court, previously stated that AI training with "transformative purposes" could cover unauthorized downloads, sparking a judicial perspective starkly different from Alsup's. It is understood that the Anthropic case will go to court on December 1, and the outcome, whether a ruling or settlement, will become a benchmark for assessing the legality of data sources in the future, while also affecting players like OpenAI facing similar disputes. Possible outcome estimations: The Trump administration, in order to maintain America's competitiveness in the AI field, may lean towards loosening the fair use scope; however, even if the court adopts a compromise route with a settlement in the billions, it would compel the entire industry to reassess its data supply chain. For the capital market, evaluating the long-term value of AI companies can no longer rely solely on algorithm breakthroughs or parameter scale but must also incorporate data legality, potential litigation costs, and settlement reserves into the model. Publishers and data platforms that can provide cleanly authorized content may hold greater bargaining power in the new order. The Anthropic copyright lawsuit is not merely a corporate dispute but resembles a mirror reflecting the tense balance between innovation and intellectual property rights. In the future, AI companies that can stand firm must not only lead in technology but also adhere to increasingly strict data governance boundaries. Related reports: Imagine RobotFi: What new gameplay do robots bring to the blockchain? Vitalik's "meow" about robots sparks heated discussion, Ethereum community: I bet all my money on those who learn to meow. Computing power reigns supreme! Jen-Hsun Huang discusses how AI can reshape global value chains, when robots will become popular, and how AI can accelerate the return of manufacturing to the U.S. "Anthropic uses 7 million books to train Claude, facing trillion-dollar piracy lawsuit! The mad rush of AI giants and the boundaries of law" was first published in BlockTempo, the most influential blockchain news media.