You may remember that we told you this more than a few months ago OpenAI has been condemned by a large group of authors. Among the plaintiffs were well-known names such as George RR Martin, Douglas Preston, Michael Connelly and John Grisham, but also The Authors Guilt, which represents more than 13,000 writers with published work, so de facto we can say that there are more than 13,000 authors directly or indirectly facing OpenAI in this lawsuit.

The basis of demand is, of course that OpenAI used their work to train its generative models. This is nothing new, hand in hand with the recent proliferation of generative models of any type of content we have seen an exponential increase in creator interest, both in the use of their creations in training and the possibility that these models can a posteriori reproduce their personal style across generations .

Probably The sector in which we have seen the most movement in this sense is music.. A few months ago, we saw an artificial intelligence create a song that imitated the work of Drake and The Weeknd, with the subsequent anger of said authors, and much more recently, this month, we saw a similar case with Bad Bunny. Now, this problem actually extends to text, images, and potentially any other generative task that is based on a generative model trained on human creations.

Well, just over two months after we first learned about the process, The hearings for the authors’ lawsuit against OpenAI started today, a case in which the authors are seeking up to $150,000 in damages from the tech company for each infringed work, among other damages, according to ABC News. The plaintiffs also requested a trial.

As we told you before, when OpenAI introduced DALL-E 3, the latest version of its generative image model, it confirmed that one of its novelties was that Authors could request that their creations be excluded from model training, something that is not present in GPT-4, but all indications are that it will be included in the future GPT-5. However, this lawsuit could force a desirable change in strategy in this regard, namely that the company must be authorized in advance to use any copyrighted content in training its models. Of course, something like this can be extremely expensive to manage, but it’s probably the only model many creators are willing to accept.