Tidal stopped paying wholly AI tracks: what a video score is

Does Tidal still pay for AI music? Not for tracks that are wholly AI-generated. Press coverage of the policy (read 2026-10-04) says that since July 15, 2026 Tidal has stopped paying royalties on music where, in its definition, every component of the track was made with generative AI. Those tracks stay in the catalog with an AI badge but cannot be monetized or sold direct to fans, and artists who use AI tools alongside human work are left alone.

That policy is about releases on a streaming service. A music bed under your own video is a different job. If you generate a bed with the Sume Music Router and lay it under a clip, you are making a soundtrack for a video, not delivering a track to Tidal. Where the two meet is when someone wants to release the bed as a song, and the rest of this post is about keeping that choice open.

What the policy says, and what it does not

Two press reports (TechTimes and EDM.com) describe the same shape. The line is drawn at wholly AI-generated, an external detection partner flags such tracks, and Tidal expects distributors to identify AI content before it arrives. I did not find a Tidal-owned page in the search results, so treat the details as reported, not as Tidal's own wording.

• Affected: tracks reported as wholly AI-generated.
• Effect: AI badge, no monetization, no direct-to-fan sales.
• Not affected, as reported: artists who use AI tools alongside human work.
• Burden: distributors are expected to flag AI content upstream.

What this means for a soundtrack you generate

For a video bed, the practical question is whether the audio will ever be released as a standalone track. If it will not, the policy does not touch it. If it might, keep the human part of the work real: write the arrangement brief yourself, record or replace a part, and keep your own notes of what was made how. Sume does not label, watermark-check or detect AI audio, and its docs make no claim about how any platform will classify a track, so that record is yours to keep.

The Music Router docs say sume/music-auto picks the engine (Lyria 3.5 today) and every Music Router model charges the fixed Music price per audio generation; the Music 1.0 page lists $0.125 per accepted generation.

Generate a bed for a video

One call, synchronous for up to 30 seconds. The docs say to put exclusions in the positive prompt and to steer length in the prompt, because duration is rejected.

import os
import requests

r = requests.post(
    "https://api.sume.com/v1/music-router/generate",
    headers={
        "Authorization": "Bearer " + os.environ["SUME_API_KEY"],
        "Idempotency-Key": "tidal-bed-001",
    },
    json={
        "model": "sume/music-auto",
        "prompt": "Warm lo-fi hip hop, 84 BPM, C minor. Dusty Rhodes chords, brushed boom-bap drums. A 30-second track. Instrumental, no vocals.",
        "mode": "sync",
        "wait_timeout_seconds": 30,
    },
    timeout=60,
)
print(r.status_code)
print(r.json())

Checklist before you publish

• Decide first if the bed will stay inside the video or become a release.
• Write the brief yourself and keep it with the project.
• Read the audio artifact from result.artifacts[] where type is audio, and use the media.sume.com URL, not a provider URL.
• Check each platform's current audio rules on its own page before you upload.