Unreal borders and copyright

I have often spoken about the new European Community guidelines on Artificial Intelligence, considering that we will also soon have the sanctions and policies of the judiciary bodies. But, like any new "stake," care must be taken to ensure a free market, competition between large and small players, and the user impacted by the service or product.

If this does not happen, they will be just another limitation on the competitiveness of an entire industry and continent because, in a connected world, geographic borders are increasingly random and ephemeral. This is, unfortunately, already happening today with unreal boundaries and copyright.

Without going into complex details, let us look at Matisse's work called "The Dance." This work is in the Hermitage Museum in St. Petersburg today. This is a famous work, but who can legally exploit its image for commercial purposes without paying attention to the copyright? It depends.

It depends on where the company that wants to use it is located and where it intends to distribute the derivative product. Copyright law differs from industry to industry. Every year, sites specializing in literature publish works that have become public domain, according to their country's calculation. In 2021, for example, the writings of Cesare Pavese slowly became accessible in Italy. The United States, on the other hand, will have to wait until 2036 to access its first novel.

Italian law stipulates that copyrights expire 70 years after the artist's death. U.S. law provides that rights expire after 95 years from the creation/publication of the work. Matisse's work has been in the free domain in the United States since 2005, having been created in 1910. Italy will have to wait until 2024, Matisse having died in 1954. China is adopting the same principle as Italy, but lowering the waiting time to 50 years, as is Canada.

A law is a vision of the world one wants to build, but it needs not be an insurmountable limit to the possibility of change, hopefully for the better (but defining what "better" means requires a separate book). On the other hand, it will also be necessary to realize quickly if something is not working: avoiding a repeat of the prohibition season.

Will we have different AI models for each country in the future for ethical regulations, GDPR, and copyrights of works used to train them? Will we continue to legislate even for AI for unreal boundaries as we did for copyright? What are the alternatives?