GenAI’s copyright issues driving model diversity

This week’s edition of the Economist (subscribers only) ran a feature on artificial intelligence and copyright. Generative AIs have been and continue to be trained on copyrighted material, including texts, images, music, videos, and more.

Not all creators are amused. Some have chosen to sue the companies developing these Generative AI models. Others, including Associated Press, Axel Springer, and Le Monde, have established licensing agreements with these companies, granting them permission to use their content for model training. If this trend persists, we might see a more varied landscape within the Generative AI industry, with models differing greatly in terms of the content they’re based on and their capabilities. This shift might also signal the end of truly free open-source models.

Meanwhile, a third group of media companies, such as Getty Images and Bloomberg, are developing their own proprietary Generative AI models using their content. Like the authors of the referenced article, I question whether these models can compete with larger, more general models. However, they may become the only way to access certain copyrighted material.

Court decisions and legislative measures will also play key roles, potentially leading to different outcomes in different jurisdictions. In the United States, where most of the world’s most valuable Generative AI companies are based, there may be a bias towards protecting these companies. On the other hand, the European Union may be more inclined to favor copyright holders.

In summary, the issue of copyright in the Generative AI sector introduces a significant degree of uncertainty and could drastically alter the competitive landscape in the near future.

Here is an earlier blog post we wrote about


Leave a Reply

Your email address will not be published. Required fields are marked *