Generative AI: A Threat or Weapon for Original IP Owners?

One of the funnest outcomes to date of the generative AI craze has undoubtedly been anonymous internet users making songs “by” their favorite artists purely using AI software. My favorite one so far has been this Kanye deep fake. And it’s not just Kanye, the same has been done for Drake, Rihanna, and basically every other big name artist.

In all seriousness, this isn’t just fun. Anonymous internet users remixing original IP is a core sub-theme under the $100B Creator Economy mega-theme. After all, what is TikTok other than anonymous users remixing and adding their own flare to existing original IP?

Predictably, the artists themselves haven’t reacted as jovially as me. Drake said this was the “final straw” for him. And Universal Music Group urged music streaming platforms to take down or block any AI-generated songs.

In this reaction, the industry is repeating one of its biggest mistakes in the past decade. But it’s not too late. Can the industry learn from past mistakes and win the day this time around? Only time will tell.

The YouTube case study

YouTube launched in February 2005 and, later that year, was blessed by a certain SNL skit by the name of Lazy Sunday. The parody song was quickly pirated & distributed on YouTube… and has racked up 10s of millions of views since that fateful night.

A few years later, Viacom sued YouTube for $1 billion stating “brazen and massive copyright infringement”. The main outcome of this lawsuit was that Google (who owned YouTube by this time) agreed to add a fingerprinting technology whereby it could identify the original IP owner for any clip and properly compensate them as such.

On the surface, this was a huge win: media companies were “fairly” compensated for their original IP. But let’s consider:

• YouTube is thought to be worth over $100B within Google. The original IP owners may be compensated for their IP, but they own 0% of that enterprise value, while Google owns 100%, even though it was the original IP which made YouTube’s success possible in the first place. Seems like a missed opportunity.

• Whereas they started by trying to remove original IP, top brands now pay YouTube for distribution of their original IP through both brand advertising and paid reach. One could easily ask the question why are we paying them and not the other way around?

Yes, media companies found an OK solution to being compensated for their original IP when it appears elsewhere. It is certainly better than the rest of the internet where piracy pervades with no compensation to the original rights holder. But the missed opportunity in seeking this narrow outcome is enormous.

The bottom line: when your only goal is to protect the value of your existing IP in the short-term, it is likely that you will sacrifice a lot of future value in the long-term.

Lesson #1: The flaw with attempting to prevent duplication

In the first place, it’s nearly impossible to prevent duplication in today’s world — and it will be even more impossible given generative AI. When it is not only free to distribute high quality content but also create it (with new, improving AI tools), the number of these remixes will be nearly infinite. In his post on the topic, Ben Thompson tackles this issue adeptly with the idea of Zero Trust Authenticity.

If there will be infinite copies of an artist’s work, is the appropriate response to try to prevent each marginal copy?

Lesson #2: It’s additive not competitive

Initially, Viacom and its peers viewed YouTube as a threat to its core product, TV. The thinking went: if they’re watching it on YouTube, then they wont watch it on TV. And while this fear was well-founded in some ways (cable TV subscription has precipitously declined primarily as a result of digital video options like YouTube), it sort of misses the point.

• First: this was, in many ways, an inevitability. The game was always to shift business model to the new paradigm rather than trying to prevent it from happening. The music industry, because of the cartel-like nature of its labels, has been able to do this well with Spotify, etc — but the video industry has lagged woefully behind.

• Second: it turns out that, in many ways, YouTube was additive rather than competitive. Today, every major IP owner voluntarily distributes (and even pays for the privilege to do so) their content on YouTube. It’s rightly thought of as a massive marketing funnel rather than a competitive threat.

Lesson #3: If you can’t beat ‘em, join ‘em

What we’ve learned so far is:

• This sort of “piracy” is inevitable & infinite given the internet and, now, AI. Policing it is a futile endeavor (outside of criminal situations such defamation of character for which there already exist laws)

• Step #1 is to work with distributors to capture value from that piracy. If it’s going to happen, better to work with Spotify than Napster.

• But the media industry can go even further and take a leading stance rather than a kicking-and-screaming one. The benefits there are that they may capture some of the $100B+ of YouTube enterprise value, rather than ceding it all to Google.

• The benefit for the actual artists is even greater: there will be artists born by leaning into this phenomenon just like the generation of YouTube-turned-mainstream stars.

Major success looks like this: a reliable & holistic ability to identify & compensate original IP holders for any remixes of that IP (the fingerprinting solution). And, ownership in the platforms which rise up to enable creation & distribution of all this new, remixed content.

Partial ownership of these platforms has an additional benefit: the ability to set loose guidelines on truth. If you have a seat at the table when a song or interview is remixed or created, you can prevent defamation or certain beyond-the-pail remixing at the source vs. reactively. This could be a huge additional benefit that goes well beyond the precarious deal between rights holders and YouTube that we have today.

3 startup ideas to capitalize on this trend (other than the obvious)

The obvious one is to build the next YouTube or TikTok using user-generated AI-derived music / video remixes as the primary content medium — duh. This will undoubtedly exist in some form – it will face enormous headwinds from IP owners – and it will succeed nonetheless due to consumer demand.

So let’s get to a few less obvious, more esoteric, and (to some) more interesting ideas:

1. Fingerprinting technology: It’s great that Google was able to build a content identifier for its YouTube services. But what about all the non-Google owned properties where this content appears? Shouldn’t the identification service be omnichannel?

2. Fan fiction: How about pre-negotiating the rights to original IP in a certain vertical (let’s see negotiating song rights with labels) and allowing a set of creator tools to remix on top of those paid-for rights? Services like WattPad already exist to allow fan creation; this would build on an already massive trend.

3. 21st century A&R: Just like the DJs who quickly mastered new software in the 2010s, the ability to use new tools to remix original IP via AI will become a sought after skillset in itself. Just like in every media paradigm before, the agents who can find these hidden gems before they blow up can participate in that rise to stardom. And there are ways to automate this too. See Jukin Media, which has made a business of programmatically identifying viral user-generated content, buying the rights, and monetizing.

The bottom line is that the Media industry can choose whether to proactively jump on the generative AI train – or be dragged, kicking & screaming, when the flood gates inevitably open. The train has left the station on this one, so normative questions about what should happen are moot at this point.