EDM producers turn detectives to expose AI-generated tracks passed off as human

Producers are policing the EDM scene

A loose group of electronic musicians has started working as unofficial investigators, publicly accusing other artists of releasing songs made with generative AI while presenting them as human work, The Verge reports. The most visible of them is Max "H4RRIS" Harris, a 26-year-old EDM producer from Maine who posts videos singling out tracks he believes were algorithmically produced.

Harris told The Verge that he sees AI-generated music as "a kind of decoy art form," arguing that the people making it are chasing profit rather than trying to express anything. He contrasts that with his own workflow, which he describes as hundreds of deliberate creative decisions about sound, arrangement and mixing, made in tools such as Ableton Live, Serum and Kontakt.

The fingerprints of generated music

According to The Verge, Harris and others have built informal forensic habits for spotting machine-made songs. Harris points to a sharp hiss running through many AI tracks, which he links to the way generative models build audio out of noise and reference data. He also describes a quirk he associates with Suno in particular: vocals and melodic elements starting to stutter at exactly the same time, which he says happens because the models struggle to separate the parts of the songs they were trained on and end up treating them as one instrument.

Visuals can give tracks away too. Harris cited an AI-driven evangelical persona called Lionsaddle, whose hands flicker in and out of existence in its videos, and noted an overly glossy look to clips from the Christian house act Midnite Manna.

Suno comes up again and again. Harris credits the platform's characteristic flaws with teaching him what to look for, while fellow producer Nihil Young blames it for the flood of generated tracks now reaching the scene. Young, a 39-year-old Italian producer whose Threads posts inspired Harris's videos, claims some users upload copyrighted recordings — including Madonna songs — and have Suno remix them. The Verge points to a "Like a Prayer" cover by Josh Fawaz that now carries AI credits following a wave of online backlash.

Accusations without proof

The callouts are far from conclusive. Harris has flagged MANSA's "Midnight on My Mind" and "Take Me (To The Moon)" by Danny and Ian Asher as likely AI productions, but as The Verge notes, none of those artists has publicly addressed the claims, no definitive evidence has emerged, and Harris may simply be wrong. The accusations themselves are a symptom of a wider problem: with no reliable way to verify how a track was made, listeners and artists are defaulting to suspicion.

The cost of speaking out

Young told The Verge that after his Threads posts he received hacking attempts and a wave of harassment from AI defenders, and he believes someone bought fake followers for his Spotify profile to make his own numbers look fraudulent. He has since paused the callouts. The financial pressure is real as well: Young, who previously did mixing and mastering work for Sony Music, Warner and Universal, says he lost many of his production clients soon after generative tools arrived.

Why it matters

This is what enforcement looks like when generative audio outruns detection. Volunteers with trained ears and video channels fill the gap left by platforms and regulators, and their judgments are unverified and potentially wrong. The people making the accusations pay a price — harassment, security risks and reputational traps — while the artists they target face unproven allegations. Meanwhile the underlying economic damage, from losing clients to competing for attention against cheap, effectively infinite generated output, falls on human musicians regardless of who gets exposed. Other creative fields now absorbing generated text, images and video may soon confront the same mix of fraud, vigilante policing and eroded trust.