This is not a joke.
Sarah Silverman is the lead plaintiff in a class action lawsuit filed against OpenAI alleging that the company used her books and other copyright protected materials to train the Artificial Intelligence that fuels ChatGPT. The comedian is also a party to a similar lawsuit filed against META over META AI.
I recently realized that I too had (likely) helped train AI without when I found out that OpenAI had (allegedly) scraped blogs hosted on WordPress. I maintain a blog on Wordpress for archival purposes because a few of the posts continue to get significant traffic 15 years after I originally posted them.
In the United States, copyright owners have certain exclusive rights to protect their interests in the works. The rights for most* copyright owners include the right to make copies, perform the work publicly and make derivative works.
Here, when copyright protected works are used to train AI they are doing more than "reading" the work. The software is ingesting it to develop and support predictive language skills.
The problem here is not just that countless (and undisclosed) works have been used in this manner without permission from, or compensation to, the owners of the copyright in the underlying works. Rather, people who use tools like #chatgpt and #metaai are going to create works that rely on the preexisting works to create new works, many of which could be a derivative of the original. For example, if I ask ChatGPT to write me a science fiction screenplay in the voice of George Lucas... where do you think it will look to make the predictions?
I look forward to following these cases, and the others that have been filed, to see how the courts (and parties) are going to handle the issue of infringement by humans who developed the engines and tools that created ChatGPT and META AI.
What do you think?
*believe it or not, the owners of copyright in recorded music do not have the exclusive right to perform music unless its over satellite or digital platforms.