A staggering $1.5 billion settlement has been reached by Anthropic, an AI company, aimed at authors and publishers whose works were utilized to train its language model, Claude. This decision followed a judge’s finding that Anthropic had pirated books for training purposes. Under the settlement, authors could receive a minimum of $3,000 per book. As an author with eight published works alongside my wife’s five, the financial implications are compelling.
While significant, the settlement does not resolve the greater ethical debate surrounding whether AI firms should utilize copyrighted material without compensation. This case has shifted the discussion from theoretical to financial. It begs the question: Is it fair for AI companies to create trillion-dollar ventures without compensating the authors whose works they rely on?
Having grappled with this issue myself, I now firmly believe that authors deserve payment for their contributions. As a member of the Authors Guild, which is involved in litigation against companies like OpenAI and Microsoft for similar reasons, I recognize the complexities at play—though I speak solely for myself in this reflection.
Historically, I had been ambivalent about whether companies could lawfully train models on personal works. The notion that AI could compile a vast trove of human knowledge was appealing, as echoed by artists like Grimes, who see contributions to this evolving landscape as potentially life-extending. However, the situation becomes markedly different when considering how corporate giants utilize those works. Books, with their depth and breadth of knowledge, are foundational to the proper functioning of language models; they should not be exploited without fair compensation.
The staggering sums reported recently by tech giants at a White House tech dinner reveal the vast resources available to these companies—sums far exceeding the $1.5 billion settlement. Yet, they argue that paying authors is impractical under the existing copyright laws designed before the internet era. Moreover, legal concepts like "fair use" could favor these companies, allowing them to exploit works without direct payment.
Despite claims that establishing a fair payment system for authors is complicated, such systems exist in other creative industries like music. Mary Rasenberger, the CEO of the Authors Guild, asserts that a collective licensing system could be viable, yet AI companies seem reluctant to agree as it might challenge their claims of fair use.
Addressing this issue is fraught with new arguments, including claims that a national AI strategy must overshadow concerns about creative rights, as competing with China in AI technology is framed as a matter of national security. Critics assert that this diminishes the value of authors and their contributions, with some proposing extreme measures to bypass copyright entirely.
Even with payments forthcoming for authors, this does not solve the more pressing issue—not enough people are reading. In the tech landscape, leaders like Jack Dorsey and Elon Musk have made dismissive comments about intellectual property that reflect a broader disregard. As authors, while we may benefit financially from situations like Anthropic’s settlement, the deep-rooted cultural attitude toward books and thoughtful narrative remains under threat.
The shift in discourse from courthouse debates to discussions of fair compensation is significant, but the future remains uncertain as we grapple with the essential value of literature in an increasingly AI-driven world.
