So your books have been pirated to train AI …

As a writer, I find AI an intensely bad thing. Yeah, it’s momentarily distracting and amusing to be scrolling through social media and see what are obviously AI-generated images of a horrible, horrible person licking the feet of an equally horrible person, or even to see some fanboy’s imagining of what a Justice League movie would look like if it were made in the 1960s.

Then you realize that this is AI and valuable natural resources are being used to run servers that create these images. Not to mention that real, actual artists – and in the case of the written word, writers – could be put out of work by this.

I first had some foreboding realizations about the effect AI might have on my work a few months ago when I went looking for one of my pieces for CrimeReads, so I could post a link to it, and realized that Google AI had generated bullet points of my articles. Why would someone need to click through to CrimeReads when they could just read the AI interpretation of what the site’s writers had written?

I was aware that some writers were saying they believed entire books of theirs had been used for AI training.

I was concerned about that because I know a lot of writers. I thought no one would possibly pirate and upload my little true crime books. Who would need that?

Then, on March 20, the Atlantic published a story about the 2-million-plus books, articles and scientific papers that have been added to Library Genesis, or LibGen, which is what Wikipedia calls a “shadow library” of file-shared work, including work that is not available digitally.

One of my writer friends said a couple of her books were there. Another had 15 of her 19 books pirated on LibGen.

The Atlantic offered a real public service that allowed readers to search to see if their work, or the work of someone they know, was uploaded to LibGen for AI training.

Here’s a screen shot of my search results using the link in the Atlantic.

I found two of the four true crime books I co-wrote with Douglas Walker on there, using the Atlantic’s search engine.

I later found what purports to be LibGen’s own search portal and could not find these two books. Had they been taken down in the meantime? Was there some mistake? It seems hard to imagine that the Atlantic got that wrong. Based on that article, I saw dozens of writers, some of whom I know, posted that they had also found their books on the site.

It’s unclear what to think about what’s there and what’s not, but there’s no question that pirated work hurts writers and publishers who might not be able to sell copies of books if people can get them for free. Since shit flows downhill, that trickles down to harm for writers, that’s for sure.

We’re all still figuring this out. It’s been pretty clear for a while that AI-generated art and writing is bad for the planet – servers use a lot of water to cool to create AI – and bad for writers. I suspect it’s also bad for consumers, but then I was never one to snap up pirated books and art and have been pretty skeptical of that inclination.

Brave new world, hell. This seems like a very cowardly ploy.

Leave a comment