GenAI companies should be celebrating. They love books: all of those well-written, proofed, edited words are just perfect for training their auto-writing machines.
The snag is perhaps in the second part of the day’s official title, ‘copyright day’. GenAI companies don’t seem to like that part very much, having apparently used thousands, maybe millions, of copyrighted books, from pirate sources, to train their tools. Copyright simply asks that a user seek permission from a rightsholder and then agree to pay a fee for the use of a copyrighted work. The GenAI companies chose to ignore this reasonable process.
On this World Book and Copyright Day we call on governments to recognise the urgency of upholding copyright law and ensuring that authors and other rightsholders can enforce their rights to choose when their works are used.
How? We need immediate transparency and compliance by AI companies in obtaining copyright owners’permission through licensing and offering adequate payment as the only means to secure the legal use of books or other literary works. At the moment these works are being used to train and produce outputs at a massive commercial scale for the sole profit of generative AI companies. In addition, numerous court cases in various jurisdictions are seeking redress for copyright infringement arising from unlicensed training, with lawsuits recently brought in France, Canada and the USA against Meta’s use of a pirated book repository astraining data. It is all too obvious that the AI behemoths have been scraping illegal copies of works to boost their bottom lines.
World Book and Copyright Day links back to some of the greatest human writers the world has known. It is vital that technological progress does not entail devaluing human creation by giving unfair competitive advantages to the machine-generated content that is currently flooding the market.
Happy World Book and Copyright Day!