There’s a current debate raging within the music industry. I am, of course, talking about the recent dustup surrounding the incursion of Artificial Intelligence (AI) into the creative process that produces sound recordings. But this is not the first time recording companies and artists have sought to protect their creations from technological infringements. For example:
Let’s return for a moment to the musical battlefields of 1999. A full quarter century ago, the battle lines were drawn between content creators and owners on one side and a startup named Napster on the other. To illustrate the situation, let’s use a simple analogy.
Think of the record companies as vendors selling apples they’ve grown (CDs) from their apple cart at a price that was based, in part, on historical prices of the product. Now, along comes Napster, which sets up its apple cart right next to that of the record companies, but it operates with an unheard-of and “unique” business model: It gives away its apples (CDs) for free.
That’s right. Napster prices its apples at zero dollars and zero cents. Consumers pay nothing and still get the apples. To make matters worse, Napster had stolen the apples it was giving away from the record companies’ cart. Bad Kitty, indeed (if you remember the Napster logo). Not to mention, that’s a tough, if not impossible, business model to compete against. So, of course, litigation ensued.
At this time, while the industry is furiously litigating, along comes a company named, yes, Apple, which sees an opportunity amidst this hubbub. What did Apple do? It licensed the sound recordings for legitimate distribution and told their vast army of music consumers who wanted individual songs, “Here, you can have them for 99 cents each!” Everybody gets an apple, and everybody gets paid. Brilliant.
So, what takeaways from this chapter of musical history can be applied to today’s AI controversy? There are two:
1. Stealing copyrighted music is not a good business model.
2. Licensing content can be a good business model where all stakeholders can make money. Even the reformed Napster (Good Kitty) recognized this at the turn of the 21st century.
Now, let’s return to today’s new era of generative music. That is, new music is generated using Artificial Intelligence algorithms that “train” machine learning platforms to generate “new” music. They do this by inputting existing sound recordings, the vast majority of which are copyrighted.
(Important disclaimer: I did not attend law school or pass the bar. Nothing in this article should be relied upon or construed as legal advice.
Of course, it could be argued that while both Napster and AI did (dare I say) upset the musical apple cart, they did so in very different ways. This begs the question: Is the present-day debate over generative music similar to or different from the bad old days of outright music piracy? Which brings us to our topic:
When Is Fair Use…Fair?
That is, on today’s musical battlefield, does training machine learning platforms to produce “new” music with copyrighted sound recordings fall under the doctrine of Fair Use? Let’s begin our answer with…
The four factors that determine Fair Use:
1. The purpose and character of the use, including whether it’s of a commercial nature or for nonprofit educational purposes
2. The nature of the copyrighted work
3. The amount of the copyrighted work used in relation to the copyrighted work as a whole
4. The effect of the use upon the potential market for, or value of, the copyrighted work.
Now that we have the four factors, let’s further explore Fair Use with two more analogies:
Let’s say someone reads every physics book in the world and becomes an expert on physics. Then, they write a book on physics using all their training and knowledge. As long as the new book does not lift verbatim sections of the books read by the author and present them as new material, one could argue that the author has adhered to the Fair Use doctrine.
Another example: Let’s say someone trains their machine learning platform on every copyrighted sound recording ever created. The AI algorithm generates a new sound recording based on what it has “learned” from those sound recordings. Is the new song transformative and original, or is it a rip-off or sound-alike of an existing artist’s or sound recording? Does the new amalgam of existing music adhere to the Fair Use doctrine?
The purpose of this article is not to definitively come down one way or the other on what does or does not constitute Fair Use. That will most likely be decided by the courts. Rather, the purpose here is to highlight the fact that there are basically two paths forward for the music industry and those using copyrighted sound recordings to create generative music:
1. The Rocky Road, or…
2. The EZ Button.
The Rocky Road involves costly and time-consuming litigation, which is already occurring. This means that litigating attorneys will have no problem meeting their quota of billable hours.
The EZ Button is the industry implementing a licensing methodology that is fair to all stakeholders.
The good news is that the wheel doesn’t need to be reinvented. Licensing methodology already exists between record labels and sound recording owners with Digital Service Providers like Apple, YouTube, Spotify, Pandora, Napster, etc., and as well as cable companies. Both models use a form of blanket licensing.
In other words, this isn’t a time in musical history to lean back, grab the popcorn, and watch what transpires. Instead, this is the time to lean forward into the issue. It is a time for those concerned with AI infringement to work with legal and other professionals, including our firm, to find unpaid or underpaid royalties due.
Regardless of which side of the issue you support, we at Prager Metis greatly value your opinion. So, to help ensure that no one is stealing apples from your cart, I encourage you to contact me at jrust@pragermetis.com.
