Suno shares plans to combat spammy AI music

**Suno** is moving to curb spammy AI-generated tracks by pairing a new **watermarking** system with a tighter **download policy**. In a lengthy blog post, CEO and co-founder **Mikey Shulman** set out the company’s principles and next steps as it tries to look more legitimate to platforms, rights holders, and listeners who have grown wary of low-effort AI music flooding feeds and libraries.

On the product side, the plan is straightforward even if the internals are not fully public yet: watermark generated output so AI tracks can be identified after they leave the app, and change how downloads work so bulk export and redistribution of spammy material is harder. Together, those moves treat spam less as a content-moderation afterthought and more as a distribution and provenance problem—tag the audio, then throttle the easy paths that turn generation into scale.

For engineers and builders shipping generative media, the lesson is that quality and safety controls increasingly live at the **artifact and export** layer, not only in prompts or model filters. If your product can produce infinite low-cost output, you will eventually need durable signals (watermarks, metadata, rate limits, download rules) that survive outside your UI. Suno’s framing also shows how “transparency” is being productized: make origin detectable so downstream platforms can filter, rank, or label AI audio without guessing.

That push sits in a crowded AI music market where volume is cheap and attention is not. Platforms and labels already face floods of synthetic tracks that dilute catalogs, game recommendation systems, and muddy attribution. A watermark-plus-policy package is Suno’s bid to look like a responsible generator rather than a spam factory—useful if it wants partnerships, distribution, and regulatory breathing room while rivals race on model quality alone.

What to watch next is whether the watermarking and download rules actually reduce spam in the wild, and whether Shulman’s principles turn into concrete product defaults users cannot casually bypass. Builders should note the pattern: announce principles, ship provenance tooling, then tighten export. If you run a generative product, plan for the same sequence before platforms or regulators force it.