Bandcamp Banned AI Music. What Counts as Human Enough Now

On January 13, 2026, Bandcamp posted a short statement on Reddit titled Keeping Bandcamp Human, and with it became the first major music platform to ban AI music outright. Not label it, not demote it, ban it. I’ve sold music on Bandcamp for years and I’ve spent the months since that announcement watching artists, AI hobbyists, and label folks argue about one phrase in the policy, so let me walk you through what it actually says, how it’s being enforced, and where the line between a tool and a ghostwriter now sits.

The Short Answer

Here it is in plain terms. Bandcamp now prohibits music and audio that is generated wholly or in substantial part by AI, and it separately bans using AI tools to impersonate other artists or their styles. There’s no percentage threshold, no checklist, no appeal form spelled out in the announcement. According to Music Business Worldwide’s coverage of the policy, the company framed it as wanting musicians to keep making music and wanting fans to trust that what they buy on Bandcamp came from human beings. Enforcement leans on community reports plus Bandcamp’s stated right to remove anything it merely suspects of being AI generated.

So if your workflow is typing a prompt into Suno or Udio and uploading the render, that music no longer belongs on Bandcamp. If you play, sing, write, and arrange your own material and use AI somewhere in the margins, say for a mastering assistant or a stem cleanup pass, you’re almost certainly fine. The murky territory is everything in between, and that’s where this article spends most of its time.

What Bandcamp Actually Said

The policy itself is remarkably brief, which is both its strength and its problem. Billboard reported that Bandcamp is the first online music listening platform to prohibit both full and substantial AI use in music, and that the company asked users to flag suspected AI content through its existing reporting tools so staff can review it.

Three things in the announcement matter more than everything else. First, the operative phrase is “wholly or in substantial part,” which covers far more than pure prompt output. Second, the impersonation clause extends existing rules about copying artists into the AI era, so a model trained to mimic a specific singer’s voice or a band’s signature sound is banned regardless of how much human effort went into the rest of the track. Third, Bandcamp explicitly reserved the right to pull music on suspicion alone. That last point deserves your attention whether you use AI or not, because suspicion is not proof, and no platform on earth has a detection system accurate enough to make suspicion reliable.

It’s worth remembering why Bandcamp of all places drew this line. The platform’s entire pitch has always been direct fan support. When someone spends ten dollars on a record there, most of it goes to the artist, and the company had recently launched a thirteen dollar a month human curated subscription built around DJs and journalists picking records instead of algorithms. Music Business Worldwide noted that the ban landed about four months after that subscription launched. A store built on the promise that your money reaches a person cannot quietly fill up with music no person made. The ban is brand protection as much as it is principle, and honestly, I don’t think that makes it less sincere.

Where the Line Actually Sits

Here’s the uncomfortable truth: nobody outside Bandcamp knows exactly where “substantial part” begins, and I suspect nobody inside Bandcamp could give you a number either. The policy is deliberately written as a standard, not a rule. Standards give a platform flexibility. They also leave artists guessing.

Based on the policy text, the reporting since January, and how similar language has been enforced elsewhere, here’s the spectrum as I read it, from safest to most exposed.

Clearly fine: AI used as a utility inside a human production. Noise reduction, stem separation, a mastering service, chord suggestions you then play yourself, AI transcription of your own melodies. These tools shape audio a human created. Nobody serious is arguing this violates the policy, and Ars Technica’s coverage of the announcement made the same point, noting the policy targets the full automation end of the spectrum while leaving room for artists who fold AI tools into a larger creative process.

Probably fine but document everything: a track where you wrote and performed the core material and used AI for one contained element, like generating a texture you then chopped, resampled, and buried in a mix full of your own playing. The finished recording is substantially yours. The AI contribution is an ingredient, not the meal.

Genuinely risky: an AI generated backing track with your real vocal on top, or an AI composition you re-recorded with virtual instruments without meaningfully changing the writing. The composition, which is the heart of the work, still came from a model. The analysis published by Jack Righteous lands where I do on this: light mastering or post-processing on an AI render doesn’t change what the track fundamentally is, and only significant human rebuilding, meaning replaced elements, rearranged structure, newly recorded vocals or instruments, and a final version that no longer resembles the original render, has a real argument for compliance. Even then, enforcement is subjective and you’re betting on a reviewer’s judgment.

Banned, full stop: prompt in, finished song out, upload. Also banned at any level of human involvement: anything that imitates a specific real artist’s voice or style using AI.

Notice what test I’m actually applying there, because I think it’s the one that matters. Don’t ask how many hours you spent. Ask who authored the composition and the performance. If the answer to both is “a model,” no amount of curation, tag writing, or cover art effort makes the music human enough for this policy. Curation is real work. It just isn’t the work Bandcamp is protecting.

How Enforcement Really Works

Bandcamp did not announce a detection system, and as far as anyone has reported, it doesn’t run one. This matters more than most coverage admits. An analysis by Undetectr, a site that, full disclosure, sells services to AI music creators and therefore has its own angle, pointed out that Bandcamp had historically been the easiest platform for AI uploads precisely because it ran no metadata scanning or spectral analysis. The new regime replaces nothing with a human process: fans and artists report suspicious releases, staff review them, and Bandcamp removes what it judges to be in violation.

That design has two predictable consequences, and both showed up within weeks of the announcement.

First, some AI creators lost catalogs with no warning. Side-Line’s coverage of the community reaction quoted a user who said their transparently labeled AI music, disclosed as AI generated and human curated right on the album pages, was removed without notice, and who argued a transparency requirement would have been fairer than a blanket ban. Another commenter in the same thread accepted the need to stop the flood of automated junk but objected that intentional, serious AI assisted work got swept away with it. Whatever you think of AI music, the deleted without notice part should give every artist pause. Keep your masters backed up somewhere you control. Bandcamp is a storefront, not an archive, and January proved it.

Second, human artists will get falsely flagged. It hasn’t become a documented epidemic yet, but the mechanism guarantees it eventually. Report buttons plus a suspicion standard plus genres that sound synthetic by design, think vaporwave, dungeon synth, certain strains of ambient and hyperpop, equals wrongful takedowns. Deezer commissioned a survey in November 2025 and found that 97 percent of participants could not reliably tell AI generated songs from human made ones. If ordinary listeners can’t tell, ordinary reporters can’t either. If you make electronic music with quantized drums and synthetic vocals, your best insurance is a paper trail: project files, session recordings, stems with timestamps, even quick phone videos of you working. Boring, yes. Cheaper than fighting a takedown with nothing, also yes.

The Numbers That Forced Everyone’s Hand

If the ban feels sudden, the data behind it doesn’t. The volume of machine generated music entering the market over the last eighteen months is genuinely hard to overstate.

Deezer, the French streaming service that built its own AI detection tool, has published the clearest running tally. When the tool launched in January 2025, roughly 10,000 fully AI generated tracks were arriving per day. By September 2025 it was 30,000. By January 2026, around 60,000 daily, or 39 percent of everything delivered. By April, nearly 75,000 a day, which is more than two million AI tracks a month. By July 2026, Deezer reported close to 90,000 fully AI generated tracks per day, crossing half of all daily uploads on peak days, with over 13.4 million AI tracks detected and tagged in 2025 alone.

Now the part of Deezer’s data that I think explains the platform bans better than any ideology: almost nobody is listening to this stuff. Fully AI generated music accounts for only about 1 to 3 percent of total streams on Deezer, and the company found that up to 85 percent of the streams those tracks do get are fraudulent, meaning bot driven plays designed to siphon royalties, which Deezer demonetizes. As TechCrunch summarized, tagged AI tracks are also stripped out of Deezer’s algorithmic and editorial recommendations, and the company has begun licensing its detection tech to the rest of the industry. Deezer even stopped storing high resolution versions of AI tracks to save server costs. Read that again: the majority of new “music” entering streaming is content almost no human chooses to hear, uploaded mostly to farm fraudulent royalties.

Spotify saw the same flood and chose management over prohibition. In September 2025 the company said it had removed more than 75 million spammy tracks in the previous twelve months, and it rolled out three measures: a stronger impersonation policy covering unauthorized AI voice clones, a spam filter targeting mass uploads and artificially short tracks, and AI disclosures in credits built on DDEX, an industry metadata standard, so artists can specify whether AI touched vocals, instrumentation, or post production rather than facing a crude AI or not AI toggle. Spotify was explicit that it was not banning AI music, just the fraud and deception around it.

And then radio moved. In late November 2025, iHeartMedia launched a program called Guaranteed Human, with programming chief Tom Poleman telling staff the company would not use AI personalities or play AI music with synthetic vocalists posing as human. Billboard’s report on the program noted DJs now fold the Guaranteed Human line into their hourly legal station IDs, and that the policy pulled Xania Monet, an AI voiced act built by poet Telisha Jones using Suno and signed to a multimillion dollar deal with Hallwood Media, off iHeart’s airwaves. Billboard also reported around the same time that Suno alone was generating about seven million tracks a day. A smaller rock and metal streaming platform called ROKK had already banned fully AI generated music back in July 2025. Bandcamp wasn’t a lone crank. It was the sharpest edge of a trend.

How Bandcamp Compares to Everyone Else

If you’re trying to figure out where a given track can legitimately live in mid 2026, here’s the landscape as my own releases and my research see it.

Bandcamp: banned if AI did the substantial creative work. Human review, community reports, removal on suspicion. No disclosure option exists that makes substantially AI music acceptable.

Spotify: allowed with conditions. No AI ban, but the spam filter, impersonation rules, and DDEX disclosures mean undisclosed slop and voice clones get removed while disclosed, rights clean AI music can stay.

Deezer: allowed but quarantined. Detected AI tracks get tagged for listeners, excluded from recommendations and editorial playlists, and fraudulent streams are demonetized. Deezer has also said it will remove AI tracks with no streams in six months.

iHeartRadio: banned from broadcast under Guaranteed Human, along with AI DJs and AI podcasts.

The distributors, the companies that actually deliver your files to streaming services, are where things get practical, and they disagree with each other more than the platforms do. The comparison published by Undetectr matches what I’ve seen artists experience: DistroKid is the most permissive of the big three, using an AI disclosure checkbox at upload and focusing enforcement on fraud, impersonation, and spam behavior rather than AI as a category. TuneCore sits in the middle, collecting more granular AI attribution in its metadata fields. CD Baby holds the strictest line, rejecting fully AI generated content outright, and reporting through 2026 indicates it updated its terms to allow terminating accounts tied to fully AI music. A comparison by MUSICΔ made a point I’d underline: most distribution enforcement targets behavior, meaning mass uploads, bot streams, impersonation, and thirty second royalty bait, and an honest artist releasing genuine original work rarely hits trouble.

For current pricing context, since costs come up the moment anyone talks distribution: DistroKid’s Musician plan runs $24.99 a year for unlimited uploads after a 2026 price rise, with Musician Plus at $44.99, per a 2026 comparison by Ones to Watch. On the generation side, Suno’s Free tier gives 50 daily credits for personal, non commercial use only, while Pro costs $10 a month, or $8 a month billed annually, with 2,500 monthly credits and commercial rights, and Premier runs $30 a month, or $24 annually, with 10,000 credits and the Suno Studio workstation. Note that commercial rights attach to songs made while subscribed; free tier output can’t be retroactively monetized by upgrading later, and Suno has signaled that download limits and new licensed models are coming, which could reshape those terms again.

If You Make Music With AI Tools

First, the hard part: if your catalog is substantially AI generated, Bandcamp is gone for you, and pretending otherwise by scrubbing tags or hoping nobody reports you is a bad plan on a platform where removal requires only suspicion. You’d be building a store that can vanish overnight. I’d also gently push back on services promising to make AI tracks undetectable. On Bandcamp there’s no detector to beat, only human reviewers, and everywhere else you’d be undermining the disclosure systems that are the actual path to legitimacy for AI assisted music. Spotify’s DDEX disclosures exist precisely so responsible AI use doesn’t have to hide.

Where should substantially AI work live instead? Direct sales channels without blanket AI bans: Patreon, Ko-fi, Buy Me a Coffee, or your own store on Shopify or Wix, all of which the Jack Righteous guide flags as viable, provided you follow their rights rules and, in Patreon’s case, its specific AI content policies. On streaming, DistroKid with honest disclosure remains the widest door. Just go in clear eyed about Deezer style quarantining: tagged, untouched by recommendations, and earning almost nothing organically, because the 1 to 3 percent listening share tells you the demand side hasn’t shown up.

If you’re a hybrid artist, someone who genuinely writes and performs but uses generative tools somewhere meaningful in the chain, your job is twofold. Keep the human authorship at the center of the composition and performance, and keep receipts: project files, stems, drafts, session dates. On Bandcamp specifically, I’d also describe your process in the album notes. Reviewers responding to a report are making a judgment call, and a credible, specific account of how a record was made is worth more than silence.

If You’re a Fully Human Artist

Mostly, this policy is good news for you, and the honest thing is to say so. Your Bandcamp page just became one of the only storefronts on the internet that certifies, however imperfectly, that fans are buying from people. That’s now a marketing asset. Use it.

Your two action items are defensive. Back up everything off platform, because January showed Bandcamp will delete first and let people complain later. And build the habit of keeping work in progress evidence, especially if your music leans electronic or heavily produced. The Deezer finding that 97 percent of listeners can’t distinguish AI from human music cuts both ways: it means your very human record can sound like a machine to a stranger with a report button. When a takedown lands on the wrong artist, and it will happen to someone, the artists who recover fastest will be the ones who can hand over a folder of session files the same day.

Four Questions to Ask Before You Upload Anything to Bandcamp

When artists ask me whether a specific track is safe under the new policy, I run them through the same four questions. They won’t give you certainty, because the policy itself doesn’t offer any, but they’ll tell you which side of the line you’re probably on.

Who wrote the composition? Not who had the idea, who wrote it. If the melody, chords, and structure came out of a generation model and you kept them, the composition is AI authored no matter how much you fussed with the arrangement afterward. If you wrote it and merely used AI to sketch, suggest, or transcribe along the way, it’s yours.

Who performed it? A real vocal take, real playing, or even carefully programmed parts you built note by note all count as human performance in my book. A rendered stem you didn’t shape at the note level doesn’t. Tracks fail this question more often than the first one, because it’s so tempting to keep an AI backing track that already sounds finished.

Could I prove the answers to a stranger in ten minutes? This is the enforcement reality check. A reviewer looking at a reported release doesn’t know you. Project files, dated stems, and process notes in the album description are how a stranger decides you’re credible. If you can’t show your work, assume the benefit of the doubt goes to the reporter, not to you.

Would I be comfortable if the process were printed on the cover? This one is less about Bandcamp and more about you. The disclosure systems Spotify is building around DDEX are pushing the whole industry toward a world where the process is on the cover, or at least in the credits. If honest labeling would embarrass the release, the release has a problem that no platform policy workaround will fix.

Run a track through those four and most of the gray area shrinks. A song that passes the first two comfortably is almost certainly fine under any reasonable reading of substantial part. A song that fails both is banned however you dress it up. And a song that splits them, human on one, machine on the other, is exactly the kind of release where I’d keep documentation, disclose in the notes, and accept that a removal is possible even if you believe you’re compliant.

My Honest Take

I think Bandcamp made the right call for Bandcamp, and I think a lot of the commentary about it, on both sides, is lazy. The pro ban crowd tends to pretend the line is obvious. It isn’t; “substantial part” is a judgment standard enforced by tips and vibes, and it will produce unfair outcomes at the edges. The anti ban crowd tends to pretend this is a censorship story. It isn’t that either; it’s a specialty store deciding what it sells, the same way a farmers market can decline to stock instant noodles, and the numbers show the flood it’s defending against is mostly royalty fraud, not misunderstood art.

The prediction I’ll put my name on: within a year or two, human verified becomes a standard product tier across music, the way organic became one in food, complete with labels, premiums, and arguments about certification. iHeart trademarking Guaranteed Human and Deezer licensing its detection tool to the industry are the early signs. Where does that leave the question in this article’s title? Roughly here: you are human enough for Bandcamp when a human authored the composition and the performance, and AI, if present at all, worked as a tool inside that process rather than as the process. Everything else about the Bandcamp AI music ban, the reports, the reviews, the removals, is just machinery for approximating that sentence.

One Thing to Do This Week

Whatever kind of artist you are, spend one hour on this: download your full catalog and project files to a local drive plus one cloud service you control, then write three sentences in each Bandcamp album description about how the record was actually made, who played what, where, with which tools. It protects you if you’re wrongly flagged, it’s honest disclosure if you use AI anywhere, and in a market drowning in ninety thousand anonymous machine tracks a day, a specific human story attached to human made music is quietly becoming the strongest sales pitch there is.

Sources

• Music Business Worldwide, Bandcamp bans AI-generated music: ‘We want musicians to keep making music’: https://www.musicbusinessworldwide.com/bandcamp-bans-ai-generated-music-we-want-musicians-to-keep-making-music/
• Billboard, Bandcamp Bans AI-Generated Songs in a First for Major Online Music Platforms: https://www.billboard.com/pro/bandcamp-bans-ai-generated-music-songs-policy/
• Side-Line, Bandcamp bans AI music uploads with new generative AI policy: https://www.side-line.com/bandcamp-bans-ai-music-new-generative-ai-policy/
• Hypebot, Bandcamp Bans AI Music: https://www.hypebot.com/bandcamp-bans-ai-music/
• Jack Righteous, Bandcamp Bans AI Music: Where Creators Can Sell in 2026: https://jackrighteous.com/en-us/blogs/ai-music-distribution-guide/bandcamp-ai-music-ban-where-to-sell-2026
• Undetectr, Bandcamp Banned AI Music: Where to Distribute Your Tracks Instead: https://undetectr.com/blog/bandcamp-ai-music-ban-2026
• Music Business Worldwide, Spotify has deleted 75m+ tracks in ‘spammy’ AI music crackdown: https://www.musicbusinessworldwide.com/spotify-has-deleted-75m-spammy-tracks-as-it-unveils-new-ai-music-policies/
• Music Ally, Spotify reveals its latest measures to handle AI music, spam and deepfakes: https://musically.com/2025/09/25/spotify-reveals-its-latest-measures-to-handle-ai-music/
• Deezer Newsroom, AI-generated tracks now represent 44% of all new uploaded music: https://newsroom-deezer.com/2026/04/ai-generated-tracks-represent-44-of-new-uploaded-music/
• Deezer Newsroom, AI Music Tops 50% of Daily Uploads on Deezer: https://newsroom-deezer.com/2026/07/ai-music-exceeds-50-percent-daily-uploads-deezer/
• TechCrunch, Music streamer Deezer says more than 50% of daily uploads are AI-generated: https://techcrunch.com/2026/07/21/music-streamer-deezer-says-more-than-50-of-daily-uploads-are-ai-generated/
• Billboard, iHeartRadio Bans AI Music, Podcasts & Radio DJs With New Program: https://www.billboard.com/pro/iheartradio-bans-ai-music-podcasts-radio-djs-new-program/
• Undetectr, DistroKid vs TuneCore vs CD Baby: AI Music Policies Compared (2026): https://undetectr.com/blog/distrokid-tunecore-cdbaby-ai-policies-compared
• MUSICΔ, AI Music Distribution Terms Compared: DistroKid, TuneCore, and CD Baby: https://musicdelta.com/en/articles/ai-music-distribution-terms
• Ones to Watch, DistroKid vs TuneCore vs CD Baby: Best Music Distribution: https://resources.onestowatch.com/distrokid-tunecore-cdbaby-comparison-2026/
• Tech Jack Solutions, Suno Pricing: Complete Guide to Free, Pro & Premier (2026): https://techjacksolutions.com/ai-tools/suno/suno-pricing/