You probably haven’t read Between the Devil and the Deep Blue Sea: The mission to rescue the hostages the world forgot. It’s the story of how a former British diplomat rescues a group of sailors who spent five years hijacked by Somali pirates. Sadly, it wasn’t quite the bestseller that its author hoped it might be –which I know, because I wrote it, and thus far it’s only sold a few thousand copies.
Over in California, though, there is one enthusiastic reader who has seemingly ploughed through the book from end to end and absorbed every word. No, not some Hollywood mogul who wants to make a Netflix series out of it, but a miserable bugger who hasn’t even bothered buying a copy, despite being worth about $1trillion. For my book, along with about six million others, is among those that appear to have been used by Mark Zuckerberg, the plutocrat boss of Meta/Facebook, to train his new generation of artificial intelligence products.
As you’ll be aware if you read March’s edition of The Atlantic magazine – and should be aware of if you’ve ever written for the BJR – Meta has used LibGen, a vast pirated library containing more than seven million books, to train its new flagship AI model, Llama 3. The magazine gleaned this from Meta documents disclosed in a copyright infringement lawsuit brought by multiple writers in the US. The documents appear to show that Meta knew it might be legally dodgy to use LibGen but pressed ahead anyway.
The lawsuit, whose plaintiffs include several Pulitzer-prize winners and the comedienne Sarah Silverman, is one of a dozen copyright test cases now pending in the US courts, with famous authors such as John Grisham and Jonathan Franzen also involved. But for the benefit of lesser writers such as me, The Atlantic also devised a handy online search tool so that we can check whether our books too are among LibGen’s terabytes of plundered print. Sure enough, when I typed my name in, my book came up, as did several articles I’ve written for the BJR, which is among thousands of academic journals pirated by LibGen. If you try The Atlantic’s search tool, you’ll probably see that some of your own contributions are there too.
LibGen, for those who’ve never heard of it, was set up around 2008 in Russia, purportedly so that people in former Soviet countries and the developing world could read academic research and books for free. Its creators say they’re providing a service to folks who can’t afford costly academic
journal subscriptions or to download books like mine (£7.47 on Kindle, seeing as you ask). So far, so very right on. But you don’t have to be a social justice warrior to see the difference between a pirate website illegally sharing stuff for free, and Meta using that stuff to build AI products to make it even richer than it is. Not only are we authors not getting paid a penny, the AI products might also end up doing us out of a job. AI-generated books are already flooding genrefiction markets, for example, and while most are pretty crummy at the moment, next-generation AI products might change that.
Meta’s response has been a “fair use” defence – that because AI transforms the LibGen book data into new, processed information, rather than simply replicating it, there’s no copyright infringement. A similar analogy might be a reporter writing a story based mainly on cuttings. But what Meta is doing is the absolute mother of all cuts jobs – and again, most newspaper databases aren’t free (Factiva, for example, costs nearly £5,000 a year).
The Society of Authors (SOA) is now urging the Government to take action against Meta in Britain over its use of LibGen, and wants the culture secretary Lisa Nandy to order its executives to explain themselves before Parliament. As the SOA chair Vanessa Fox O’Loughlin, herself a crime novelist, remarked: “A book can take a year or longer to write. Meta has stolen books so that their AI can reproduce creative content, potentially putting these same authors out of business.”
A year to write? Hmm. Maybe, if you’re doing fiction. But non-fiction books like mine – the sort that we journalists often specialise in, because of our reporting skills – can often take a year just to research. The cost isn’t just in time either. Between The Devil, for example, involved research trips to India, Thailand and the Philippines to interview the freed sailors, plus considerable costs in translators. I’d spent about £5,000 of my own money before I even had a book proposal to tout around and have only just recouped that in terms of revenues from the book (sadly, very few publishers pay research expenses in advance these days).
True, most authors write for the passion of it, not the money – which is just as well, given that average author earnings are only £7,000 a year, according to the SOA. The reason I wrote Between the Devil was to highlight the plight of sailors who had been by abandoned by the world, and the extraordinary heroism of a British diplomat in rescuing them. That I chose to spend two years labouring effectively for free, though, should not entitle Meta to take the fruits of my labour and profit from it. Especially when they haven’t even asked permission.