Context Window 75
It was a quieter week for product releases, but there were significant legal, policy and research developments. I am in Portland, Oregon next week for IBPA Publishing University: I look forward to seeing some of you there. In the meantime, have a good weekend.
The big news this week was five major publishers—Hachette, Macmillan, Cengage, Elsevier and McGraw Hill—suing Meta and its CEO Mark Zuckerberg for copyright infringement over its AI training. They were joined by author (and former Authors Guild president) Scott Turow. The plaintiffs are seeking class-action status on behalf of a wider group of copyright holders, including owners of registered copyrights in books and journal articles allegedly used by Meta.
If the publishers are successful, the lesson from Bartz v. Anthropic is that the definition of which works are eligible for the class will be critical, as that will determine who might benefit from any eventual damages or settlement. The Authors Alliance breaks this down: works registered with the Copyright Office either within five years of publication and before Meta’s use, or within three months of publication, and with a specific wrinkle for Open Access/Creative Commons-licensed work.
The Creative Industries Policy and Evidence Centre ran a good online seminar this week presenting research from Australia on author sentiment around AI. The results were very consistent with last year’s University of Cambridge report on AI and the novel, showing a high degree of concern about the impact of AI on livelihoods and reluctance to engage in licensing. You can see a recording of the presentation here.
I wrote up my reflections on the seminar, highlighting that the experience of academics may differ somewhat from trade authors. That distinction matters because attitudes to AI licensing are likely to follow the underlying incentives of each publishing market: who gets paid, who bears the risk, and what authors believe publication is for.
For wider context on the impact of AI on creative jobs, Gallup just published new research in the Journal of Cultural Economics suggesting that the main impact so far is not on wages or creative output, but on the way work is organised, with many creatives using AI as part of their practice.
It’s the third week that I’ve featured something about the rising cost of AI tools, but it’s worth reading this analysis because it speaks to one of the most important near-term challenges for heavy AI users: what happened to pricing last month was a warning shot. Thanks to subscriber Alex Boden for sharing that piece.
A really interesting use case from news publisher Semafor, which used AI to take transcripts and videos from a five-day conference and produce synthesis and analysis at a speed that would have been impossible with traditional journalism. The technology behind this was prototyped using OpenAI’s Codex and then developed further in Claude Code.
This isn’t a completely new idea, of course. Back in the distant past of digital publishing (2009), I remember being really impressed that the organisers of the Bookcamp conference put together a hardback book of notes and photos within weeks of the event—and the fact it’s still on my shelf says a lot about the persistence of physical media. Better print-on-demand and Semafor-type AI curation means that every conference or event could be a book.
I’ve written before about the AI-generated podcast company Inception Point, but there are some staggering figures in this piece: Inception now has over 10,000 shows, a quarter of them created in the last three weeks. And in the wider podcast market, nearly 40% of new feeds are for AI-generated shows. With that signal-to-noise ratio, discovery is becoming harder—and the only thing keeping this from happening to books is that the economic incentives are stronger for podcasts.
Underlining the explosion in AI audio, Spotify this week announced Personal Podcasts, a tool that lets AI agents create bespoke audio briefings from a user’s data and content, and save them directly into that user’s Spotify library.
If every event you attend, and every document you work on, can become a publishable thing through AI, then publishers face even more competition for consumers’ time.
This was originally published in my email newsletter. To receive weekly updates on how AI is affecting the publishing industry, sign up here.