Music AI's Licensing Revolution (and Other Discoveries)

Writing this from my kitchen table where on the side I have 4 batches of Makgeolli brewing. There's something poetic about the intoxicating sweet aroma of alcohol production while reflecting on current AI news. Both involve transformation, patience, and a bit of controlled chaos. I'm sure you feel it too, but everything moves so fast nowadays. Case in point: some of the biggest developments I'm about to share literally happened a few days ago.

While thinking through the current music AI landscape, a phrase kept coming to mind: "licensing is the new moat." Yesterday, ElevenLabs proved this insight right in the most dramatic way possible.

They launched Eleven Music with something their competitors don't have: actual licensed training data. While Suno and Udio are drowning in billion-dollar copyright lawsuits from major labels, ElevenLabs went ahead and struck deals with Kobalt Music Group and the Merlin Network. It's the difference between asking permission and asking forgiveness. In 2025, permission is winning.

Here's what's actually happening under the hood: ElevenLabs uses diffusion models and neural networks trained on vast, licensed music datasets. Basically, this means they can generate studio-quality music from text prompts without the legal sword of Damocles hanging over every output. The platform includes guardrails preventing users from generating songs with specific artist names or copyrighted lyrics.

The industry response has been fascinating to watch. There's genuine positive sentiment around their approach, which tells you everything about how the music world feels about AI companies that actually respect intellectual property. It's not just about avoiding lawsuits; it's about building sustainable relationships with the creative community.

This shift from "move fast and break things" to "move thoughtfully and build bridges" feels like a maturation point for AI in creative industries. The wild west phase is ending, and the companies that survive will be the ones that figured out how to play nice with existing creative ecosystems.

Speaking of playing nice, I've been hands-deep in some fascinating new tools that aren't just AI bolted onto traditional software—they're designed AI-first from the ground up.

MACE by Tensorpunk has had me experimenting for hours. At its core, it's a sampler plugin, but instead of loading existing samples, it generates them using Stability AI's Stable Audio Open model. Think of it like having an infinite sample library that creates unique sounds on demand, locally on your machine. No cloud, no API calls, no monthly subscriptions.

I'll be honest. I've been thinking about building something exactly like this for years, and I'm genuinely thrilled someone executed it so well. The workflow is seamless: you describe what you want, it generates up to 47 seconds of stereo audio at 44.1kHz, and you can immediately drag it into your DAW. It's perfect for sound design or when you need samples that literally no one else has.

Mozart AI represents another fascinating paradigm shift. It's a web-based DAW where each track has its own AI agent generating MIDI or audio content. You can also prompt a lead agent to create full loops across all tracks. Based on my conversations with the founder, they're still iterating on user experience and quality, but the agentic approach feels like the future of music production software.

The pattern here is clear: we're moving from traditional software with AI features to AI-native tools with traditional interfaces. It's not about replacing human creativity. It's about creating new collaborative possibilities.

Suno V4.5+ continues this trend with its stemming feature, which I've been using constantly. You can generate a song and then separate it into individual tracks for further processing in your DAW. The outputs are genuinely impressive. Creative, human-like, with interesting arrangements and solos that feel intentional rather than algorithmic.

But perhaps the most exciting and terrifying development is Google's Genie 3, also announced this week. Imagine having an AI serve as the game engine for open-world games. Not just generating assets or NPCs, but creating the entire interactive environment in real-time.

Genie 3 generates interactive 3D game worlds at 720p and 24 frames per second from simple text prompts. You can walk around these environments, interact with objects, and the AI maintains visual consistency across frames. This was a major limitation in previous versions.

As someone who's spent countless hours in open-world RPGs, there's something both thrilling and unsettling about this. The technical achievement is remarkable, but it also represents a fundamental shift from AI as a content creation tool to AI as an experience creation platform.

The implications extend far beyond gaming. We're looking at a future where AI doesn't just generate static content but creates dynamic, responsive environments that adapt to human interaction in real-time. It's the difference between asking AI to write you a story and asking AI to become the world where your story unfolds.

Here's something that's been working incredibly well for me lately: using AI to help me prompt other AI systems. Specifically, I've been using Claude to generate better prompts for Suno.

I created a Claude Project with a custom prompt for music generation. When I want to replicate a particular band or song style, I start a new conversation within the project, input the band name or song reference, and Claude outputs both a style prompt and an "exclude styles" prompt optimized for Suno (works for Udio as well).

## **Objective:**
Create a **concise music model prompt** for an artist that stays within **1000 characters**. Output the main prompt and exclude list in separate code blocks for easy copying into AI music model interfaces.
---
## **Search Instructions:**
**Continue searching until you find sufficient information about:**
1. **Primary Genres** (e.g., Indie Pop, Electro-Pop, Lo-Fi, Synthwave)
2. **Key Influences** (e.g., 80s/90s pop, electronic, R&B artists)
3. **Sound Characteristics** (e.g., smooth synths, warm basslines, production techniques)
4. **Mood/Emotion** (e.g., nostalgic, playful, introspective)
5. **Signature Elements** (unique instrumentation, vocal style, arrangement choices)
6. **Cultural Context** (when applicable - e.g., K-Pop influence, British sound)
**Note:** If you cannot find comprehensive information initially, continue researching using different search terms and sources until you have enough detail to create a specific, accurate prompt.
---
## **Prioritization Hierarchy (when approaching character limit):**
1. **Sound Elements** (genres, influences, characteristics) - *highest priority*
2. **Mood/Emotion** - *high priority*
3. **Signature Sounds** - *medium priority*
4. **Cultural Context** - *trim first if needed*
---
## **Output Format:**
### Main Prompt:
```
[Comma-separated descriptors under 1000 characters]
```
### Exclude List:
```
[Comma-separated terms to avoid]
```
---
## **Example Output:**
### Main Prompt:
```
Indie Pop, Electro-Pop, Synthwave, Lo-Fi, R&B, City Pop, 80s Pop, 90s Pop, Nostalgic Vibes, Smooth Synths, Catchy Melodies, Playful Energy, Introspective Lyrics, Sophisticated Production, Soft Vocals, Warm Basslines, Dreamy Soundscapes, Urban Aesthetic, Retro Influences, Chillwave, Funky Grooves, Soulful Beats, Minimalist Arrangement, Atmospheric Pads, Emotional Depth, Melodic Hooks, Shimmering Synths, Analog Warmth, Jazzy Chords, Groovy Bass, Soulful Pop, Acoustic-Electronic Fusion, Soft Inviting Tone, Vinyl-like Warmth, Hypnagogic Dreamscapes, Vintage Soundscapes
```
### Exclude List:
```
Hard Rock, Metal, Punk, Experimental Noise, Industrial, Techno, EDM, Trap, Rap, Dubstep, Aggressive Vocals, Screaming, Hyperpop, Hardstyle, Hardcore, Dancehall, Reggaeton, Loud Percussion, Fast BPM, Commercial Pop, Top 40 Radio Sound, Commercial EDM, Trap Beats, Club Music
```
---
## **Important Rules:**
- **Character Limit:** Main prompt must stay under 1000 characters
- **No Artist Names:** Do NOT include specific artist names in the output. Instead, use cultural/era contexts when needed (e.g., "Fela Kuti inspired" becomes "Political 70s Afrobeat," "Tony Allen influence" becomes "70s Nigerian polyrhythmic drumming," "Beatles-esque" becomes "60s British Invasion")
- **Edge Case:** If an artist's actual genres appear in the standard exclude list, do NOT exclude them (artist's true style takes priority)
- **Format:** Always render both outputs in separate code blocks for easy copying
- **Accuracy:** Focus on the artist's authentic sound rather than generic descriptors
---
**Below is the artist or song the user wants a prompt for:**

The results have been remarkably good. Much better than my manual attempts at prompt engineering. Yes, I'm using AI to prompt AI. This is pure prompt engineering, and honestly, it feels like the future of creative AI workflows.

This meta-approach represents an evolution in how we think about human-AI collaboration. AI systems helping optimize other AI systems. We're not just users of AI tools; we're becoming orchestrators of AI tool chains. The skill isn't just knowing how to use individual AI systems, but understanding how to make them work together effectively.

Looking across these developments, a few patterns emerge. AI tools are becoming more sophisticated and ethically aware. The companies that prioritize licensing and creator relationships are gaining competitive advantages. Human-AI collaboration is evolving beyond simple automation toward complex, multi-stage workflows.

Most importantly, we're seeing the emergence of AI-native creative paradigms rather than AI features grafted onto existing tools. This isn't just about making current workflows more efficient. It's about enabling entirely new forms of creative expression.

For those of us navigating the creative-tech intersection, the message is clear: stay experimental, stay informed, and remember that the most interesting developments often happen at the boundaries between different AI systems and human creativity.

The future isn't about AI replacing human creativity. It's about humans becoming more sophisticated orchestrators of AI capabilities. And honestly, that future is arriving faster than any of us expected 🌀

A couple of other things that caught my attention this week:

Your Favorite Strategists Favorite Strategy - A thoughtful Are.na collection that connects strategic thinking across different domains. Relevant as we think about how to strategically adopt AI tools in creative practice.

How to keep your writing weird in the age of AI - Essential reading on maintaining human uniqueness and creative voice as AI writing tools become ubiquitous. The tension between efficiency and authenticity feels increasingly relevant.

What's your experience with these tools? Are you seeing similar patterns in your creative work? I'd love to hear from others navigating this space—we're all figuring this out together.

Share