To Use or Not to Use: Generative AI and Copyright

drawn image of calendar icon

December 28, 2023

hand drawn writing icon

Anastassia Lauterbach

Would you let your classmates copy your homework? When it comes to math, there might be only one answer to an equation. But what about an essay? Maybe you put all your heart into your research and produced something unique.

Last week, I visited the Matisse Museum in Nice, France.

On 21 October 1953, Henri Matisse donated eight significant paintings to the City of Nice, with a cutout Danseuse créole (1950) that my daughter (The Girl in the Romy&Roby series) tried to copy when she was seven.

Anastassia Lauterbach at the Matisse Museum in Nice, France standing in front of of an image of Henri Matisse.

When the museum opened in 1963, these prestigious works represented the collection’s core, further enriched by legacies and donations from the painter’s family.

Matisse himself copied a lot of artists while perfecting his techniques. He used to go to the Louvre in Paris, the most famous museum in France, making copies of Raphael, Poussin, and Gustave Moreau. “Expression is everything,” Matisse used to say. He has always emphasized the necessity to make his own rules. Making our own rules is what, for me personally, distinguishes the work of a human artist from what a Generative AI app might come up with.

Anastassia Lauterbach at the Matisse Museum in Nice, France showing the importance of Generative AI and Copyright.

You might have heard about an open-source AI art generator called Stable Diffusion or Midjourney (get a beta version at https://discord.com). These tools allow anyone to create images based on text prompts.

These open-source programs are built by scraping images from the internet, often without permission and proper attribution to artists.

As a result, they are raising questions about ethics and copyrights.

According to the website Lexica, which tracks over ten million images and prompts generated at Stable Diffusion, Matisse’s name and the names Picasso and da Vinci appeared around two thousand times each.

Matisse died, and we can only guess what he might have thought about AI in paintings. Living artists say there is a risk of losing income as people use AI-generated images based on copyrighted material for commercial purposes.

And what about another question?

Think about a school essay that you really put your heart into.

Art is so closely linked to a person. It could raise data protection and privacy problems.

Berlin-based artists Holly Herndon and Mat Dryhurt are working on a tool to help artists opt out of being in a training data set.

They launched a site called Have I Been Trained, which lets artists search to see whether their work is among the 5.8 billion images in the data set used to train Stable Diffusion and Midjourney.

An industry initiative called Content Authenticity Initiative, which includes the likes of Adobe and the New York Times, is developing an open standard that would create a sort of watermark on digital content to prove its authenticity.

Nothing is without legal disputes, and the answers to the copyright questions won’t be straightforward. As an example, Google won a case against writers who objected to the company’s scraping their copyright work for Google Books.

What do you think? Should AI creators be allowed to use copyrighted work to train their models?

Use the comments below to respond and I will share the most insightful answers in my next article!

romy and roby and the secrets of sleep book cover

Book 1

Romy & Roby And the Secrets Of Sleep

Don’t forget to check out:

0 Comments

Submit a Comment

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

Select your currency