Authors and AI Companies Settle Landmark Copyright Case

Anthropic, an AI company, agreed to pay $1.5 billion in a settlement over copyright infringement claims brought by hundreds of thousands of authors. This case may not be the all-out win authors hoped for.

Plaintiffs and Settlement Details

Charles Graeber, a bestselling author and one of the primary plaintiffs, expressed mixed feelings about the lawsuit’s outcome. Although proud of standing up against Anthropic, he admits feeling the financial impact. Authors, including Graeber, are set to receive a portion of the settlement, with Graeber expected to be compensated for his works used by Anthropic’s chatbot, Claude.

A federal judge in San Francisco approved the settlement in July following claims that Anthropic trained its AI models using digitized copyrighted books without consent. The settlement amount may seem large, yet it must be divided among numerous authors and publishers. Additionally, a significant portion, over $100 million, went to legal fees.

Fair Use and Industry Implications

Anthropic asserts training AI on books falls under fair use. Although authors received compensation, the lawsuit’s limited scope didn’t rule against Anthropic’s use of copyrighted materials, provided payments are made. This outcome raises questions about broader implications, potentially benefiting AI companies.

Such fair use arguments surfaced in other cases, like Meta’s lawsuit, where authors claimed Meta used pirated copies for AI training. Meta won, citing fair use due to the authors’ inability to show market impact. AI companies often rely on fair use to justify training with copyrighted content.

International Concerns and Future Regulation

Intellectual property lawyer Dale Nelson cautioned about AI companies abroad, noting challenges in enforcement due to government control, particularly in China. Chinese AI firms use techniques like ‘AI distillation,’ which build models using outputs from other AI models, often U.S.-based ones like GPT-4 and Claude.

New York Times bestselling author Andrea Bartz, a plaintiff in the case, advocates for a non-proliferation agreement to extend regulation internationally. Despite challenges, Bartz remains optimistic about rights protection in the AI era and stresses the importance of continued legal battles.

Potential of Licensing

Licensing could offer authors a fairer opportunity. Through licensing agreements, rights holders might control their works’ appearance in AI outputs. Organizations like Authors Guild support licensing as a legal training pathway. Some AI companies have started making licensing deals with publishers.

While licensing marketplaces exist, widespread adoption is delayed due to ongoing litigation with major AI firms. Authors hope progressive steps from lawsuits will eventually create a fairer environment for creatives.