AI and the Law // BrXnd Dispatch vol. 026

AI, copyrights, trademarks, oh my!

You’re getting this email as a subscriber to the BrXnd Dispatch, a (roughly) bi-weekly email at the intersection of brands and AI. I am working on finalizing SF Conference dates and should have those in the next few weeks. As always, if you have any questions, you can reply to this email or send me a note.

Hello! Thanks for sticking around for more thoughts at the intersection of marketing and AI. Over the last few weeks, I’ve been sharing some videos from my May conference (AI intuition and hallucinations were up first). This week, I’m keeping things going with a session that was a surprise favorite for many: Copyright vs. Generative Creativity. When I told friends in the event business that I was putting my legal session at the end of the day, they all warned me it was a terrible idea. You’ve got to load the end of the day with the stuff people can’t miss, and a conversation about the law hardly ever qualifies. 

But I had some knowledge they didn’t. Mainly, I knew that nearly 75% of attendees who had told me what they were most interested in learning had mentioned the legal and copyright issues that surround this new technology. Plus, I knew I had an amazing speaker in Shannon Sorenson, the Senior Vice President of Legal and Business Affairs for the National Music Publishers’ Association. My brief for Shannon was fairly simple: assume you’re the only person in the room who has read the trademark office’s guidance around AI and help everyone wrap their minds around it. 

I loved Shannon’s talk, and I thought she offered a few really helpful frameworks for thinking about this stuff.

---

Some Quick Housekeeping

In addition to the San Francisco conference, I am working on a series of events for the fall in partnership with various brands. As part of that, I am curating a set of AI companies to come present to those companies and am looking to schedule as many product demos as possible. If you fit into this and are interested in getting in front of hundreds of hungry marketers, can you be in touch to schedule a demo? Also, if you aren’t listed in the BrXndscape: Marketing AI Landscape, be sure to request to be added.

---

I particularly liked how she split input and output—the former focused on what’s in the corpus the model is being trained on, and the latter about whether that shows up in the eventual content produced.

When it comes to copyright concerns on the infringement side, it's about the exclusive right to reproduce and distribute. If I own a copyright in a song, I have the exclusive right to reproduce it, distribute it, and authorize others to make derivatives like remixes and arrangements.

On the input side, data aggregators, web crawlers, or scrapers send bots out to scrape everything on the internet. They put it into big data repositories, and then AI is trained using that data. If this data includes copyrighted material, which it often does, it could potentially violate copyright owners' rights to reproduce and distribute their works.

This leads to the output side. If an AI generates something that includes anything copyrighted—for instance, a song that includes a snippet of lyrics or a sequence of notes found in the training data—you might end up with an infringing derivative work.

And, of course, what everyone was waiting for was a good explanation of the copyright and trademark office’s take on the “copyrightability” of AI: if a model produces some text or an image, can you copyright it? Here’s Shannon again:

The Copyright Office issued this guidance in response to a particular case where they had to reject a copyright application. In this case, someone claimed they were the author, but in reality, an AI was the author. This was a comic book case where the text was written by a human, but the images were generated by an AI. The Copyright Office granted copyright for the text and the selection and arrangement of the images, but they couldn't grant copyright for the images themselves because they weren't human-authored.

So, the guidance that came out on the heels of this case essentially states that copyright only protects human creativity and human authorship. Therefore, if something is purely AI-generated, it cannot receive copyright protection.

(If you want more, I wrote about the famous monkey selfie a few years ago.)

Finally, I thought Shannon’s conclusion about the size of this issue was well put:

This will likely be one of the big questions that we see answered over the next several years as people take cases to court or challenge the Copyright Office's determinations. The current guidance states that if you're not interacting with the material after it comes out of the AI, it's considered AI-generated. This holds true even if you put a lot of direction into the AI on the front end.

For instance, if you commission a work from a painter, you might specify a general color scheme or subject matter, but the artist is the one ultimately executing the work. Similarly, even if you provide a lot of direction to an AI, if the AI is making the final creative decisions about how the work will be executed, it's not copyrightable.

Of course, there’s much more depth in the whole conversation, so check it out here or on YouTube:

And, of course, big thanks to my friend and WITI co-pilot Colin Nagy for moderating.

That’s it for this week. I hope you enjoyed it. I’ll be sharing more videos as I get them up on YouTube. As always, feel free to be in touch if there’s anything you want to talk about or I can help with.

— Noah