AI-Generated Content Policy

Last updated: 13 May 2026

InterSpace Distribution welcomes producers and songwriters who use modern tools to make music — including artificial-intelligence (“AI”) tools — provided that the underlying creative work, the training data behind the tools and the rights position of the finished recording are clean. This policy explains what is allowed, what is not, what you must disclose to us at the point of upload, and what we will do when the policy is breached.

This policy supplements our Terms of Use, Anti-Fraud Policy and Privacy Policy. Where this policy conflicts with the Terms of Use, the Terms of Use prevail; where it conflicts with the operating rules of a digital service provider (a “DSP”) to which we deliver your release, the stricter rule applies.

1. Definitions

• AI-generated content — a master recording, instrumental, vocal, lyric, melody, arrangement, cover art or any other element of a release that has been produced, in whole or in substantial part, by a generative machine-learning system. “Substantial part” includes any case where the system supplied the core musical material that a listener would recognise as the song.
• AI-assisted content — content created primarily by a human author where AI tools have been used in a supporting role, for example: mastering and mixing assistants (LANDR, iZotope Ozone, Sonible), stem separation (RipX, Moises) used on material you legally own, melody or lyric ideation as a starting point for human rewriting, vocal pitch correction and tuning, AI-suggested chord progressions, MIDI continuation in a digital-audio workstation, or licensed sample-pack generation. Tools of this kind are permitted without restriction provided their use complies with the rest of this policy.
• “Fairly trained” — for the purposes of this policy, a generative AI model is “fairly trained” if all of the following are true:
  1. The model’s training corpus was lawfully licensed from rights-holders, created by employees of the developer, in the public domain or otherwise obtained with the demonstrable consent of rights-holders.
  2. The developer of the model can produce, on request, a description of the rights position of the training data sufficient to satisfy a DSP audit.
  3. The developer permits commercial release of outputs and does not assert a claim over your master.
  The third-party Fairly Trained L-Certificate (fairlytrained.org) is one acceptable form of evidence, but we will also accept written confirmation from the developer or a published training-data statement.
• Voice cloning — the synthesis of a vocal performance using a model trained on, or conditioned by, a recording of an identifiable real human voice.
• Deliverer — the InterSpace Distribution user who submits a release for delivery to DSPs (typically the artist, label, manager or sub-distributor).

2. What is allowed

Plan eligibility. AI-assisted content, as defined in section 1, may be released on any plan. AI-generated content may be released on the Artist Plus and Label plans only, and always with the disclosure required by section 4. Releases on the free tier or the Artiste plan that are found to contain AI-generated content are returned at review with an invitation to upgrade.

You may distribute releases through InterSpace Distribution that contain AI-generated or AI-assisted content where every tool you used satisfies the “fairly trained” criteria in section 1, you hold the rights to all output, and you make the disclosure required by section 4. Tools we currently consider compatible with this policy include, without limitation:

• Generative composition / instrumental tools that publish a training-data statement — for example Stable Audio (Stability AI), Soundraw, Beatoven.ai, Boomy, Loudly, Mubert (commercial tier), AIVA, BandLab SongStarter and similar.
• Mastering, mixing and post-production assistants — LANDR, iZotope Ozone / Neutron, Sonible, eMastered, CloudBounce and equivalents.
• Vocal-processing tools used on a vocal you legally own — Antares Auto-Tune, Melodyne, Synthesizer V (with a licensed voice database), ACE Studio, and other tools where the synthesised voice is licensed and is not the cloned voice of a real person.
• Stem-separation tools used on material you have written or to which you hold the master rights — Moises, RipX, Spleeter, Audionamix.
• Lyric, melody or arrangement ideation tools used as a starting point that you then rewrite or develop substantially — this includes general-purpose language models.

This list is illustrative. A tool not named here is not, by that fact alone, disallowed; conversely, a tool named here may be disallowed in a specific case if the way you used it breaches another part of this policy (for example, by training a voice model on a real artist without consent).

3. What is not allowed

3.1 Suno and Udio are not permitted

Releases generated in whole or in part by Suno (suno.com / suno.ai) or by Udio (udio.com) may not be distributed through InterSpace Distribution. We will reject Suno- and Udio-originated submissions at moderation and, where we discover a Suno or Udio origin after delivery, we will recall the release from DSPs, freeze any earnings attributable to it and may suspend the deliverer’s account. This rule applies to every Suno and Udio tier, including paid plans that grant commercial-use licences to the user.

The reason for naming these two tools is that neither Suno’s nor Udio’s training corpus has been demonstrated to satisfy the “fairly trained” criteria in section 1, and both companies are, at the time of writing, the subject of significant pending litigation in the United States brought by major record labels in respect of that training. Until the position is resolved and each company publishes a training-data statement we can audit, releases originating from Suno or Udio present an unacceptable rights risk both to InterSpace Distribution and to the DSPs we deliver to. We will revisit this position if and when the underlying facts change.

3.2 Other restrictions

You may not use InterSpace Distribution to deliver any release that:

• was generated, in whole or in substantial part, by any AI tool that does not satisfy the “fairly trained” criteria in section 1, regardless of brand;
• uses a voice clone of an identifiable real human being without that person’s prior written consent — this includes deceased artists, celebrities, politicians and any private individual;
• was produced by stem-separating a recording you do not own or licence, and then redistributing the resulting stems as new material;
• reproduces, in a recognisable way, the protected musical or lyrical expression of a third-party work without a sync, sample or interpolation licence, regardless of whether an AI tool was used in the process;
• imitates a specific artist’s voice, style or branding in a way likely to mislead listeners into believing the artist endorsed or recorded the track (so-called “in the style of” releases attributed to or implying the named artist);
• is one of a large batch of formulaic AI tracks delivered with the apparent purpose of farming streams, gaming algorithmic playlists or laundering listener bots — see our Anti-Fraud Policy;
• contains AI-generated cover art that itself infringes a third-party right (for example, art trained on a named illustrator’s work without licence, or art that reproduces a celebrity likeness without permission).

4. Mandatory disclosure at upload

When you submit a release through the InterSpace CMS at cms.interspacemusic.com, you must complete the AI disclosure question for every track in the release. The options are:

1. No AI used — the track was produced entirely without generative AI tools. Mastering assistants, pitch correction, stem separation on your own material and the other AI-assisted tools listed in section 2 do not need to be declared under this option.
2. AI-assisted — the track is primarily a human composition and performance, with one or more AI tools used in a supporting role. You must name each generative tool used in the disclosure field, even if the contribution was small (for example, a lyric draft from a language model, an AI-generated drum loop, an AI-generated topline you rewrote).
3. AI-generated — the core musical material of the track was produced by a generative AI tool. You must name each tool and, where the tool is not on the public list of tools we have pre-cleared, you must provide a link to or copy of the tool’s training-data statement so that we can verify it against the “fairly trained” criteria.

By submitting the release you confirm, on behalf of yourself and any co-rights-holder, that the disclosure is accurate and complete, that you hold all rights necessary to commercially exploit the track in every territory of the world, and that no part of the track was generated by Suno or by any other tool prohibited by this policy. We pass your disclosure on to DSPs in the metadata of your release where they require it (for example, Deezer’s AI-content tag and Spotify’s emerging AI-music labelling). Some DSPs may decline a release, demote it in editorial or remove it from algorithmic playlists on the basis of an AI tag — that is a DSP decision, not ours, and we cannot guarantee placement for AI-generated releases.

5. How we verify disclosures

We operate an internal AI-detection system that scores every uploaded audio file across thirteen acoustic and metadata signals (including spectral, rhythmic, harmonic and stereo-field characteristics) and compares it against a continually updated library of known AI-tool fingerprints, including Suno’s. The score informs our human moderators; it does not by itself produce a takedown decision. Where the detection score is high and the disclosure is “No AI used” we will pause the release, contact you for an explanation and may require evidence of the human authorship of the track (for example, project files, stems, video of the recording session). Repeated discrepancies between your disclosure and our detection will be treated as a breach of this policy and of our Anti-Fraud Policy.

6. Splits, credits and royalty position

An AI tool is not a featured artist, songwriter or producer for the purposes of credits. Where a track is AI-assisted you should credit the actual humans involved (top-line writer, lyricist, mixer, mastering engineer, etc.) and may, optionally, list the AI tool used in the track notes. Where a track is fully AI-generated and there is no significant human authorship, you remain the deliverer of record and the rights-holder of the recording for distribution purposes, but you may not register the work with a performing-rights organisation as a human composition unless that PRO permits AI-generated registrations. It is your responsibility to know the rules of your PRO.

7. Cover art and music videos

The rules above apply to cover artwork and to music videos delivered alongside an audio release. AI-generated cover art is permitted where the model used satisfies the “fairly trained” criteria and the output does not infringe a third-party right (including the likeness or trade dress of an identifiable artist or brand). AI-generated music videos must comply with the equivalent video policies of the destination DSP (notably YouTube’s altered-or-synthetic-content disclosure rule).

8. Enforcement

Where we have a reasonable belief that this policy has been breached we may, at our sole discretion and without prior notice:

• reject the release at moderation;
• recall a delivered release from any or all DSPs;
• place a hold on royalties earned by the release while we investigate;
• suspend, restrict or terminate the deliverer’s account on InterSpace CMS;
• retain any non-refundable fees paid for delivery, video distribution, marketing tools or other Services;
• refer the matter to the affected DSP, to the relevant rights-holder, or to law-enforcement authorities where the breach involves fraud, impersonation or large-scale stream manipulation;
• permanently bar the deliverer from creating new accounts.

Where royalties are withheld, we will release them only once we are satisfied that the breach has been resolved and that no chargeback, takedown notice or DSP recoupment claim remains outstanding in respect of the release.

9. Appeals

If your release has been rejected or recalled under this policy and you believe the decision is wrong, you may appeal by writing to legal@interspacemusic.com within thirty (30) days of the decision. Your appeal should include the release title, the InterSpace release ID, the AI tool(s) used, the tool’s training-data statement (or a link to it) and any evidence of your human authorship. We aim to acknowledge appeals within five (5) business days and to issue a final decision within twenty (20) business days.

10. Changes to this policy

The legal and commercial landscape around AI-generated music is moving quickly. We may amend this policy from time to time to reflect new DSP rules, court decisions, regulator guidance, the publication of training-data statements by previously opaque developers, or our own operational experience. When we make a material change we will update the “Last updated” date at the top, and where the change directly affects active releases we will notify the deliverer by email or in-app notice. Continued use of the Services after a change constitutes acceptance of the revised policy.

11. Contact

Questions, disclosures that don’t fit the upload form, or notices of suspected infringement under this policy should be sent to:

InterSpace Distribution Ltd — Rights & Compliance
Suite 12, DDS Complex, 24 Airport Road, Rukpokwu
Port Harcourt, Rivers 500102, Nigeria
Email: legal@interspacemusic.com · support@interspacemusic.com