Authors petition to stop Meta stealing books

I have long been against the idea of training LLMs using sources for which the company creating the product refuses to pay. I was immensely pleased to read in The Guardian early this month that a bunch of prominent British authors—including Richard Osman, Val McDermid, Tom Stoppard, Kazuo Ishiguro and many others—have written an open letter to call the bosses at Meta to the British parliament enquiring about their unlicensed use of others’ works:

If, as you say, your Government is serious about supporting authors as part of its Creative Industry Sector Plan, and its ambition to grow and protect the UK’s world-leading creative industries which contribute £126 billion a year to the UK’s GVA, it must demonstrate its commitment and stand up against the unethical and illegal practices of tech giants, which have such a devasting impact on the lives of UK authors.

It remains to be seen how much the government really values its creatives before AI technologies which contribute a paltry £1.2 billion (as of 2023). The letter follows an explosive article in The Atlantic about how Meta pirated ‘millions of books’ (and which ones) to train its model. Some of the authors even protested last Thursday before the Meta HQ in London.

Keep in mind that while Meta did not pay for these books but stole stolen ones, it will nevertheless make lots of money from the resulting LLMs it uses across its suite of appalling products.