For years now, publishers in all categories of media have been fighting a battle with AI firms to be compensated for their content that was used to train large language models. This week there was a big media headline about Anthropic agreeing to a $1.5B settlement to pay book publishers and authors for titles that were illegally used to train AI models. Publishing industry insiders are hoping this may start a wave of similar settlements from other AI companies. There are some nuances behind this ruling, though.
For one thing, it’s only going to include books that were downloaded from illegal websites that had pirated those books in the first place like LibGen (no longer online), but not for books that were purchased and then used for training (which might qualify as “fair use”). Whether you agree or not, this is a story still unfolding in real time. October 10th is the date the judge has set to publish a list of books that will be included in this payout.
Beyond the big story, though, there are other big moves that are shaping this emerging space of how and if publishers get paid licensing fees for their content. The RSL Standard & Collective, for example, allows web publishers to set terms for how AI can use their work and how they get compensated for it. In the process, they remove the argument that AI firms have no way (or terms) for compensating content creators.
If you’re a publisher, this matters a lot … but even if you’re not, this idea that any of us can set terms for how and when we want to be compensated for the things we create and post online should be something we are all invested in seeing happen. As the common online refrain goes, if you’re not paying for the product then you are the product. Getting paid when your content is used that way seems like the least we should all be asking for.