Love this newsletter? Forward it to one curious friend. They can join in one click.

Welcome back to the AI For Humans newsletter!

Did you know that people still listen to radio? This surprised me as well. But maybe not as much as the first AI artist ever to chart on a Billboard radio airplay chart.

Xania Monet isn’t real & yet people don’t seem to care.

The brainchild of poet Telisha Nikki Jones, Xania Monet isn’t new to the charts (she hit the digital charts a month ago or so) nor an unknown in the music industry.

In fact, “she” was the subject of a bidding war and landed a $3m recording contract. According to all the stories about Xania, Telisha made Xania’s music using Suno, the popular AI music tool.

AI music is one of those areas that we don’t cover as much as we once did. When Suno debuted in mid-2023, we were enthralled with prompt-to-song and kind of shocked at how good it was even back then. And it’s gotten a lot better.

But, in the AI space, it always seems to take a back seat to the latest AI video model or frontier LLM.

And now, in retrospect, it’s a bit easier to understand why.

When it’s done well, it’s much harder to tell that AI musicians are AI.

After all, we’ve been trained on a decade plus of auto-tune and pop artists that tweak their voices. And unless AI music is purposefully jokey, it doesn’t automatically come across as AI.

When you listen to Xania Monet can you tell?

Compared to a lot of the stuff on the pop charts, it is sounds eerily similar to me.

The Udio / UMG Deal + Controversy

There was another big AI music story this week:

Udio (the other, smaller AI music generator company) settled its lawsuit with Universal Music Group. Even further, they partnered with the major music label for future releases of AI models of UMG artists.

But, in addition, Udio also announced that due to the settlement its users would no longer be able to download the songs they’d generated. 

Udio users revolted and the team is now allowing 48 hours for everyone to download the tracks they made.

But this pointed to something new in the AI space: a corporate overlord that was going to police, and more importantly, monetize these efforts going forward.

Look, it’s pretty clear that these models ingested a ton of recorded music (music that is legitimately owned by these labels & the recording artists) to train their models. And we assume that both Suno & Udio knew the risk they were taking to make these tools in the first place.

But interestingly, unlike the AI video space which seems to be operating on the idea that the ‘output’ is what you police (at least for now), the music industry wants to own the means of production in large part because they own the ‘input’.

This is a pretty big dividing line across many of the big AI copyright fights going on right now. Anthropic just settled a lawsuit brought by authors for a whopping 1.5 billion dollars in part to get ahead of this input vs output conversation.

And, ICYMI, Suno is also being sued by the major labels but has decided to fight it… for now.

So… Where Does This All Go From Here?

As per usual, the music industry and recorded music leads the way in technology adoption and disruption. After all, way before we had iPhones or Spotify, we had Napster and the iPod.

And, if we look back at how those changes reverberated across the rest of the media industry, we might be able to peek into the future.

I think the Xania Monet & UMG/Udio stories both point to a different sort of creation and a different sort of creator.

We’ve been talking about the explosion of AI content for sometime (and, yes, of AI slop) but now what we’re likely to see next the is explosion of professionalized AI content.

A future where the individual creator in their room isn’t just toiling away on some indie project but that the money and machinery of traditional Hollywood will start to push these creators to equal footing with ‘traditional’ artists. And, very likely, a blurring of those categories.

In the same way that ‘Internet Creators’ arose in the last ten years to equal footing with traditional stars (and are now repped by traditional Hollywood agents and managers), I suspect AI creators and their creations will see a similar rise.

And that soon we, the collective we, will stop caring whether whichever music or show we’re watching is made by AI. In part, because everyone will be using some form of AI but more likely…

We’ll just care if we like it or not. And Hollywood will care if it sells.

- Gavin

In this week’s AI For Humans: Sora 2’s New Character Cameo + 1x Neo Blowback👇

We 💛 This: Sora 2 Character Cameos + More

I can’t stop generating videos of my dog Ollie.

Sure, it’s fun to make videos of myself and Kevin and friends but I’m astonished at how good Sora 2 is able to recreate my dog with just a few seconds of footage in their new Character Cameo tool.

One of the best things the Sora team at OpenAI has done is really lean into new features within the Sora app. They’re now letting you generate Character Cameos directly from your Sora generations, there’s a leaderboard for Users, Remixes & Cameos to see what other people are doing as well.

And now that the Sora app is available to everyone without an invite code in the US, Canada, Japan and Korea, you should go grab it & start making videos with Ollie (add him to any gen with @olliepurcell). I’ll be sure to tweet out a few of my favorites.