What ARIA actually decided

The Australian Recording Industry Association didn't issue a vague statement of intent. They shipped a code of practice with a hard rule: wholly AI-generated tracks are not eligible for the ARIA Charts. Recordings that use generative AI in a supporting role stay eligible — but the human has to be substantially there. The rule took effect on the charts debuting Friday, August 28, 2026 (Variety, Stereogum).

Concretely, ARIA now sorts tracks into three buckets — and only the first two count:

The trigger was undeniable: a fully AI-produced cover of Madonna's Like a Prayer, by Queensland DJ Josh Fawaz, hit #1 on Australian radio play and #4 on two ARIA charts in July 2026 (SBS News). One track. And the industry blinked.

Why house and electronic got hit hardest

Every genre is dealing with the AI flood, but electronic music is the softest surface. Vocals-forward pop is harder to fake convincingly — the ear catches the wrong phrasing, the wrong breath. Instrumental deep house, techno, ambient, chillstep? A cold pad, a repetitive kick, a synth arp — that's the exact grammar generative models nail first. Prompt "melancholic deep house 4AM night drive" into any Suno-clone and you get a passable 3-minute track before you finish typing.

That's why streaming platforms are drowning in it. AI generators are dumping tens of thousands of house/lo-fi/ambient tracks onto Spotify every week, many of them attributed to fake artist profiles. Decoded Magazine flagged the same pressure inside dance music specifically: the genre where "artist" is often already an alias, a persona, a mystery — is the genre where an AI-generated ghost is hardest to catch. ARIA's rule is Australia-specific, but the electronic scene worldwide has been quietly begging for exactly this line.

"The Fawaz track wasn't a warning shot. It was proof of concept — and everyone in a studio saw it."

The real debate: tool, or substitution?

This is where it gets uncomfortable, because electronic music has never won an argument about "cheating." The sampler was cheating. The drum machine was cheating. Autotune was cheating. Sidechain compression, quantization, sample packs, one-shot kits, splice, serato — every generation, some voice in the room said this isn't real music, and every time, the culture answered the tool is the instrument now. Deep house exists because someone decided the 909 wasn't cheating. Trap exists because someone decided TR-808 samples weren't cheating.

So the honest question isn't is AI cheating. It's: where does the tool end and the substitution begin? A producer using AI stem separation to isolate a vocal is still writing the track. A producer prompting a model to output a finished master and slapping their name on it is not — that's the model writing the track. ARIA's "substantially human-made" language is trying to draw exactly this line, and it's the right question even if the enforcement will be messy for years.

The distinction that matters isn't did you touch AI, it's did you make the taste decisions. Which vocal, which key change, which drop, which mix balance. The moment those decisions leave the human, the track leaves the tradition. Everything before that is just the newest tool in a long history of accused tools that ended up being instruments.

// The way this was meant to be heard

Producer gear still wins the taste fight.

If the taste decisions have to stay human, the tools that shape them still matter. My daily kit:

  • Audio-Technica ATH-M20x — honest reference for mix decisions.
  • Sony WH-CH720N — noise cancelling for late-night arrangement work.
  • JBL BassPro Hub — the low end where a house drop lives or dies.

Full night-drive setup on the Gear page.

See the Night Drive Gear →

What it changes for the independent producer in 2026

Concretely, three things:

1. Declaration becomes normal. If ARIA's framework gets copied (BPI, RIAA, SNEP have been watching closely), producers will need to declare AI-assisted use — the way food labels declare ingredients. Not to punish the tool, to make the taste-line visible. Right now most producers using AI-assisted anything (stem separation, iZotope AI mastering, Splice AI, generative sound design) don't say a word. That era ends the moment one major chart follows Australia.

2. Fake artist profiles get harder to hide. The Spotify flood of nobody-made-this house tracks was a discovery-algorithm problem. Charts adding a "wholly-AI" filter creates a real disincentive: no chart eligibility means less playlisting, less radio, less monetization at scale. The floor won't drop overnight, but the ceiling just did.

3. Human-made becomes a marketing signal, not just an ethical one. Watch for "100% human-made" badges the way we watched for "organic" on food packaging. It sounds cynical, but it's where the market is going. On this site, every drop — from Haters to Timberlake — is human-produced. Not because I'm allergic to tools, but because the taste calls (which sample to keep, which bar to cut, which vocal to feature) are the whole job.

So where do you land?

ARIA drew a line. The rest of the industry is watching. And every producer is going to have to answer the same question, probably faster than they want: what's the last taste decision you can outsource before the track stops being yours?

Every one of the old tool wars — sampler, drum machine, autotune, quantize — landed the same way: the culture eventually decided the tool was the instrument. Maybe AI does too, in five years. Maybe not. But we're going to spend the next 24 months arguing about it, and the shape of every chart, every royalty check, and every "producer" credit depends on where the line lands.

Where do you draw it? Reply on the video, argue with me in the comments, forward this to the producer who's been quietly generating stems in Suno at 3AM. Because the argument is happening whether we join it or not — and the honest producers should be the loudest voices in the room.

Sources

5 alt titles & 3 visual ideas

Alt titles (pick your fight):

Visual ideas: