Bandcamp's AI Ban, Six Months On: What It Means for Labels Like Us

On January 13, 2026, Bandcamp announced via a Reddit post from its support account, and a companion blog post titled "Keeping Bandcamp Human," that it would ban music "generated wholly or in substantial part by AI" from its platform. The policy also prohibits using AI tools to imitate another artist's style or voice, and bars AI companies from scraping Bandcamp's catalog to train models. Six months on, with enforcement running on human review rather than automated detection, the results are messier than the announcement suggested, and the stakes for a label like ours, which sells almost everything through Bandcamp, are real.

What did Bandcamp actually ban?

The policy itself is short. Bandcamp will not host audio "generated wholly or in substantial part by AI," and it treats AI-driven impersonation of another artist's sound as a separate, explicitly banned category. As reported by TechCrunch, the company framed the move around trust rather than technology, writing that it wants "musicians to keep making music, and for fans to have confidence that the music they find on Bandcamp was created by humans." Consequence added that the rules also forbid AI companies from scraping Bandcamp's music to train generative models, a provision that matters as much to labels as the upload ban does.

Crucially, Bandcamp said upfront that it would not build automated AI detection. Enforcement runs on two tracks: user flagging through the existing reporting tools, and a Bandcamp team reviewing what gets flagged. The company explicitly reserved "the right to remove any music on suspicion of being AI-generated," which is a lower bar than proof. MusicTech noted the same language, and both outlets pointed out that Bandcamp promised to keep updating the policy as the technology and the problem evolve. That flexibility sounded reasonable in January. It has been the source of most of the friction since.

How has human review actually worked, six months in?

The honest answer is unevenly. Community sentiment on the day of the announcement was overwhelmingly positive, with Reddit threads calling it "based" and describing Bandcamp as reinforcing its identity as the last direct-to-fan platform that still feels artist-run. That goodwill has not fully survived contact with enforcement.

Reports through the spring documented a pattern: a listener or a subreddit flags an artist page as suspected AI, Bandcamp's team reviews it, and in several cases the account is suspended or the release pulled with little or no advance warning. A Frankfurt-based darkwave project called A Shrine To Failure had its 2025 album pulled after a fan wiki flagged it based on AI-styled cover art and a fast release schedule, not on any technical analysis of the audio itself. A goth/darkwave artist based in Kazakhstan, Moonvampire, had an entire catalog removed after a single Reddit report; Bandcamp confirmed the takedown as a Terms of Use violation, and the artist has publicly disputed the finding, saying the tracks used purchased instrumental beats with original vocals rather than generative tools. Neither case involved a published technical audit of the audio, because Bandcamp's process was never built to produce one.

That is the tradeoff of choosing human review over automated scanning. It avoids the false-positive problems that have dogged AI-detection tools elsewhere in the industry, but it replaces algorithmic error with a different failure mode: reputation-based suspicion, decided by whoever happens to file a report first, with no published evidentiary standard for what counts as proof either way. Musicians who believe they were wrongly flagged have reported waiting one to two weeks for a response after appealing directly to Bandcamp's support channel, and some say their catalogs were gone before they had any chance to respond at all.

For a genre space like darkwave, goth, and underground electronic music, where lo-fi production, heavy processing, and unconventional vocal chains are the norm rather than the exception, the ban has had a chilling effect that goes beyond the artists actually removed. One independent label owner told reporters he shelved several finished albums he considered technically impressive and fully coherent as records simply because he judged the reputational risk of an AI accusation too high to release them at all. That is the quieter cost of a suspicion-based policy: it doesn't just remove content Bandcamp catches, it discourages content nobody would have flagged as fake to begin with.

Why Bandcamp's ownership history makes this policy matter more

None of this is happening on a platform with a stable, artist-founded owner calling the shots. Bandcamp was founded in 2007 and stayed independent for over a decade before Epic Games bought it in March 2022. Epic sold it just eighteen months later, in the fall of 2023, to Songtradr, a licensing and sync company, in a deal that came with layoffs cutting roughly half of Bandcamp's staff and a subsequent lawsuit over the transition. That is two ownership changes and one major workforce cut in under two years, all before this AI policy existed.

That history is why the AI ban reads as more than a content-moderation footnote. Bandcamp's entire value proposition to independent labels rests on trust: trust that a purchase actually pays the artist a real share, trust that the storefront will still exist next year, and trust that what a fan buys is what the artist actually made. A platform that has already changed hands twice under financial pressure does not have unlimited credibility to spend on a moderation policy that looks good in a press release but produces false positives against real, human-made underground music. Every wrongly pulled release is also a reminder to labels that their entire catalog sits on infrastructure they do not control, owned by a company whose priorities have shifted before and could shift again.

Bandcamp's outright ban versus Deezer's detect-and-demonetize approach

The contrast with Deezer is instructive. Deezer took the opposite route: rather than banning AI-generated tracks outright, it deployed an in-house AI-detection tool starting in early 2025, tagged over 13 million AI-generated tracks that year, and by January 2026 said it was demonetizing streams on fully AI-generated songs where it found evidence of fraud, reporting that as much as 85% of streaming activity on those tracks looked artificial. Deezer has also said it plans to license its detection technology to other platforms.

The two approaches optimize for different things. Deezer's model keeps AI content on the platform, labeled and stripped of most of its royalty payout, treating the problem primarily as stream-fraud economics on a platform with billions of plays a month. Bandcamp's model treats AI content as an authenticity problem on a platform built around direct sales and a much smaller, more relationship-driven audience, so it opts to remove rather than tag. Deezer's approach scales with automation but risks classification errors at volume. Bandcamp's approach avoids automated misclassification but shifts the error risk onto community reporting and staff judgment calls, with the same downside of wrongly penalizing real artists, just via a different mechanism.

What this means for independent labels like RTS.FM

For a label that sells its own releases direct to fans on Bandcamp, the policy is a net positive in principle. Nobody running an artist-first catalog wants to compete against a flood of AI-generated tracks undercutting real production time and real studio hours, and a platform willing to draw a hard line on impersonation protects labels whose sound could otherwise be cloned and resold. The scraping ban matters too, since it is a direct statement that the label's catalog is not free training data.

But the six-month record says the policy needs a published appeals process with real evidentiary standards, not just a support email and a wait. Until Bandcamp builds that, every label on the platform carries some background risk that a single bad-faith report could pull a release with no warning. The practical response is to keep clean records: stems, session files, and production notes for anything that uses processed vocals, sample packs, or unconventional synthesis, the kind of production techniques that get flagged by pattern-matching listeners even when nothing AI touched the track.

Six months in, Bandcamp's stance is still the right instinct for a platform built on direct artist-to-fan trust. Whether it stays worth that trust depends on whether human review gets an actual appeals process before more real catalogs disappear over a suspicion nobody had to prove.