Screw the money — Anthropic’s $1.5B copyright settlement sucks for writers

Anthropic has agreed to a historic $1.5 billion settlement in a class-action lawsuit with authors, marking the largest U.S. copyright payout. This resolves claims related to Anthropic's illegal acquisition of copyrighted material from 'shadow libraries' for AI training, rather than the act of training AI on copyrighted works itself. A federal judge previously ruled that training AI on copyrighted content is permissible under fair use, deeming it 'transformative.' This outcome sets a significant precedent for numerous ongoing lawsuits against major tech firms regarding AI and intellectual property, potentially solidifying the legal framework for AI development by upholding the 'transformative' nature of AI training, provided data acquisition is lawful.

Analysis

Anthropic's agreement to a historic $1.5 billion settlement in the Bartz v. Anthropic class-action lawsuit clarifies a critical legal distinction for the AI industry. The settlement resolves claims related to the illegal acquisition of copyrighted books from 'shadow libraries', not the act of using copyrighted material for training Large Language Models (LLMs). A key development from the case is the federal judge's ruling that training an AI on copyrighted works is a 'transformative' use protected under the fair use doctrine. For Anthropic, which recently raised $13 billion, the settlement represents a significant but manageable cost of business, essentially a penalty for data piracy rather than a challenge to its core operations. This outcome establishes a powerful precedent for the dozens of similar lawsuits faced by other major technology firms, including Meta (META) and Google (GOOGL), by legally separating the permissible act of AI training from the illicit act of data sourcing. Consequently, the primary legal risk for AI developers now shifts from the use of copyrighted data to the provenance and legality of its acquisition.

Key Decisions for Investors

• Investors in AI-focused companies like Meta and Google should recognize the 'transformative' fair use ruling as a significant de-risking event for the core LLM development process, potentially bolstering the long-term investment case for the sector.
• The key litigation risk for AI developers has now pivoted from the use of copyrighted training data to its acquisition method; therefore, diligence should focus on a company's data sourcing policies and potential exposure to claims of piracy or illegal procurement.
• While this settlement sets a precedent, investors should continue to monitor the outcomes of the numerous other ongoing copyright lawsuits, as a different ruling in another jurisdiction could re-introduce legal and financial uncertainty for the industry.