The Suno Executive Doesn't Speak for Me Either
A response to "At Indie Week, I Watched the Music Industry Sell Out to AI," from the other side of the line
A response to “At Indie Week, I Watched the Music Industry Sell Out to AI,” from the other side of the line
I almost didn’t read this one. The headline scans as another AI-music rant, and I make music with Suno, so I know how those go. I read it anyway, then went through it a second time on audio, because Jeffrey Anthony didn’t write a rant. He wrote the most serious thing I’ve seen on this subject, and he’s right about more of it than I wanted him to be.
The short version of his piece: he went to Indie Week to sell the industry a way to tell human music from AI music, and found out the industry had already made the question irrelevant. The execs call their artists’ work content. The Suno exec calls her machine’s output music. The pipes being built underneath both of them pay out either way.
And before I say anything else: the man can write. The running bit with his wife about this political economy, the notebook line comparing opt-in provisions to the choices Vito Corleone offered, an epilogue that actually earns its jazz club. I read a lot of AI criticism. Almost none of it is this alive.
So let me add the one thing only someone on my side of the line can add.
The Suno executive doesn’t represent me either.
The pitch and the person in it
When she said the company exists to democratize access to music creation, she was describing someone like me. 56, software engineer, forty years of songs I could hear but never finish. I’m the guy in the marketing deck.
I make music with Suno. Three albums, an EP and three singles this year under my own project, with another album stuck in distribution right now. On top of those, two more albums as commissions, made to order for people who asked me by name. I know how those numbers sound. Hold that thought, because they’re part of the story.
And the democratization is real, as far as it goes. But here’s what the panel version leaves out. The same ecosystem selling it doesn’t want my music once I’m honest about how I made it.
This month, TikTok’s own distributor rejected a track of mine on a false sample flag. I appealed with full provenance. Prompts, timestamps, links. The appeal is what killed it. Once I disclosed AI production, a policy clause appeared that exists nowhere in their 19-page license agreement, and eleven approved tracks from the same album, same tool, stopped mattering. The working rule is don’t ask, don’t tell. I wrote it up with receipts.
Jeffrey calls the new AI labels theater, disclosure without consequence, because the money routes the same either way. From where I sit it’s slightly worse. Upstream, in the rooms he visited, disclosure means nothing. Downstream, in the ticket queues, disclosure is the one thing that gets punished. The system doesn’t want the distinction legible in either direction. His theater thesis, confirmed from below.
So no, the people on that stage aren’t on my side any more than they’re on his musicians’ side. His artists are catalog. I’m growth metrics. Same room, same math.
The part where he’s right and it costs me something
His hardest claim is that an AI track has no originating act. No bodies in a room making irreversible decisions in time. The listener hears “the smoke of a performance whose fire never burned” and supplies the fire themselves.
I won’t argue with that, because on the performance, he’s right. Nobody sweated over the takes on my album. No hands played those parts. The Copyright Office says the renders aren’t authorship and I don’t argue with them either.
But look at the image at the top of his article. A calling card from 1901 where Sousa wrote out a musical phrase for a socialite. A few bars of notation. No performance attached, none ever needed. Nobody would say no musical act occurred on that card.
That’s the honest version of my claim. Not that prompting is performing. It isn’t. But formulation can live upstream of performance, in the concept, the lyric, the structure, the hundred decisions you can’t take back about what the thing is and why. I write those. Every word, every concept, the sequencing, which take survives and which forty don’t. The machine performs it, the performance is the machine’s, and I’ll say that plainly instead of hiding it.
And the framework cuts against me too, which is how I know it’s honest. On the side, I’ve demoed and polished about a hundred songs for a songwriter who works the old way, words and melody first. On those, the formulation is his. I’m the guy bringing the machine, closer to a demo studio than an author. If formulation is what counts, then his name belongs on that work, not mine. I’m fine with that. That’s the test working.
Whether that upstream act is enough to call the result music is a real argument, and I’d rather have it about what people like me actually do than about the worst prompt anyone ever typed.
A different place to draw the line
I should say where I’m standing when I read him, because it’s not neutral ground. I’ve spent many years with non-dualism and the last year or so inside A Course in Miracles specifically - a tradition that treats our pairs of opposites as lines we draw rather than the deep structure of things. So when I hit his fire and smoke, his icon and index, his act and object, I notice that every load-bearing wall in the essay is a two, and every two puts my music on the wrong side of the wall.
The twos are real at the level where he draws them. Four musicians two feet away making irreversible decisions together is a different event than a render, and I’d never tell him otherwise. That night at The Pocket happened. The wave that went through his body happened.
But his framework treats feeling as property. The fire belongs to the body that burned it, the recording transmits it, the listener receives it, and anything that triggers the response without the original ownership is counterfeit.
The tradition I sit in says feeling was never property. Meaning arises in the meeting, every time, or it doesn’t. His own Langer half-agrees: art is not the discharge of the artist’s private emotion, it’s a form, and meaning accrues to forms at the site of recognition. The fire he felt at The Pocket wasn’t shipped to him from the stage. It arose where he was sitting. That doesn’t make it less real. It makes it his.
And no, this isn’t the argument he already swatted away, the one that says provenance stops mattering as long as something makes you feel. I’m asking a different provenance question. Not which substrate made this, a body or a spreadsheet. What is this in service of.
That line cuts through the middle of both camps. An endless machine feed engineered to harvest attention and convert it to capital is in service of fear, and it fails no matter how the audio was made. A session player grinding out quantized, pitch-corrected product for a label that calls it content can fail the same test with human bodies on every track. Four masters in a basement giving twenty people an image of lived life passes completely. And forty years of not being able to finish, finally over, with my name on what came out and the people who know me listening, well, you can judge that one yourself. I know what it felt like. It felt like the opposite of fear.
Here’s the thing. His essay already demonstrates this line better than any theory could. He spent two days at that conference glancing at the back door, waiting for authorities who never came, then three weeks walking around his house unable to stop thinking about it. His words, not my diagnosis. And none of it left his body until four musicians played it out of him. The conference sections of his piece were written from dread. The Pocket section was written from love. You can hear the difference in the prose. The line he’s looking for runs exactly there, and it doesn’t run between carbon and silicon.
About the shame
He has a whole separate piece arguing that having the machine dress up your inadequacy isn’t creativity, it’s Promethean shame wearing a mask.
I lived the shame half of that essay for forty years. The first time my music played in public was a high school radio station in 1985, and I spent the next four decades embarrassed enough by the gap between what I heard inside and what I could build that almost nothing left the room. I wrote about it. That part of his diagnosis I can verify from the inside.
What I can’t verify is the second half. When the work finally met my bar, what came out wasn’t a mask, and it wasn’t old material either. The songs from those forty years are mostly gone (a PC crash in my late 20s took a decade of them, no backup). What broke loose was the ability to finish. Everything I released this year is about right now. My career and the thing changing it. A spiritual text I’m in the middle of studying. My wife’s TikTok community.
That release count up top that sounds like a content mill? A mill makes what the feed wants. Nobody’s feed asked for a trip-hop record about A Course in Miracles. And nobody commissions an album from a mill by name. Two of this year’s albums exist because Paula and Brandi each asked for one, the way you’d ask a friend who bakes. That’s not a content strategy. That’s the Sousa card again, a musical act done for a particular person, and it’s the oldest legitimate reason to make music there is.
The streams tell the same story. Twelve thousand or so this year, and I’m probably ten percent of them, because I love my music and I play it constantly. Nobody runs a mill for an audience that’s mostly himself. None of this is aimed at a saturated commercial lane. It’s the music I actually want to hear, made on nights and weekends around a full-time job, and the day it stops being that, I’ll stop.
I’m not clearing inventory. I’m finally able to keep up with my own life.
If that’s shame, it’s the strangest shame I’ve ever felt. Shame hides. This won’t shut up about its own life.
The refusal, from my side
He ends with a bifurcation. One population accepts its role as training data. Another refuses.
Minority report: some of us on the AI side refuse too. Tools exist that strip AI signatures from generated audio. They work. Using one would have gotten my rejected track approved. I didn’t, and won’t, because making AI music pass as human surrenders the claim that it’s legitimate. I disclose everywhere the form allows, and where the form punishes disclosure, I write it up in public instead.
Last week made that refusal more expensive to say out loud. Hacked Suno source code just confirmed what the lawsuits alleged: the company scraped YouTube Music, Deezer, and the stock libraries at scale, using rotating proxies to get past YouTube’s bot protections. Read that next to my choice above. I wouldn’t defeat the detectors to get one track through. The company that built my instrument defeated someone else’s to build it. So when I ask for legibility, understand that it runs upstream too, and it applies to Suno hardest of all. The executive on that panel said democratize. The source code says something else.
His CEP certifies that a recording came from embodied acts, unquantized, uncorrected, human all the way down. My receipts certify the opposite provenance, openly. Opposite certificates, same enemy: a system that profits from nobody being able to tell. And the same bet, that some listeners want to know what they’re hearing and choose it on purpose.
I believe every word of his night at The Pocket. Nothing I render will ever be four people two feet away making something irreversible together, and the thing I make isn’t trying to be. If his certification helps people find that and pay for it, good. Genuinely.
Both of us are asking for the same small thing the industry has decided not to build. Legibility. Let people see what a thing is, and trust them to care.
The backstory is in I Could Always Hear It. I Just Couldn’t Build It. The distribution receipts are in Releasing an AI-Produced Album in 2026. Everything else is at crumrine.net.