Suno AI Music Platform's Copyright Filters Found Bypassable, Raising Concerns for Artists

Suno, an AI music platform with a stated policy prohibiting the use of copyrighted material, has been found to have easily bypassable copyright filters for both audio and lyrics. This circumvention allows for the generation of AI-created imitations of copyrighted songs, which has led to concerns among artists regarding unauthorized monetization and potential royalty diversions. While streaming services acknowledge the technical challenges and report investing in detection systems, Suno has declined to comment on the matter.

Introduction

Suno, an artificial intelligence music platform, operates under a stated policy prohibiting the use of copyrighted material. The platform permits users to upload original tracks for remixing or to set original lyrics to AI-generated music, utilizing systems designed to detect and prevent the use of copyrighted content. However, reports indicate that these copyright filters can be circumvented with minimal effort.

Methods of Circumvention

Users have demonstrated methods to bypass Suno's copyright detection mechanisms for both audio tracks and lyrics.

Basic modifications to audio tracks, such as altering speed or adding white noise to the beginning and end, using free software like Audacity, have been shown to bypass filters. These modifications can subsequently be reversed within Suno Studio. Once uploaded, the modified track can serve as a base for new AI music. The company's $24-a-month Premier Plan includes access to Suno Studio, which facilitates this process.

While directly copying official lyrics often results in the system generating unintelligible vocals or flagging the text, minor spelling changes to copyrighted lyrics can bypass the detection filter. This enables the AI to generate vocals that closely mimic original artists.

Characteristics of Generated Content

AI-generated imitations of popular songs, including Beyoncé's “Freedom,” Black Sabbath’s “Paranoid,” and Aqua’s “Barbie Girl,” have been produced, showing significant resemblance to the originals. While most listeners may discern the difference, some outputs could potentially be mistaken for alternate takes.

When no style transfers are applied, Suno's model 4.5 or 4.5+ typically reproduces the original instrumental arrangement with minor sound palette adjustments. Model v5 may incorporate more distinct instrumental elements, taking greater liberties with the source material.

AI-generated vocals can closely mimic original recording artists, though they may exhibit a lack of nuance and dynamic range compared to human performances. Instrumentals in these AI covers can also simplify artistic choices made in the original works.

Implications and Artist Vulnerability

The creation of unauthorized covers through these methods violates Suno's stated purpose and terms of service.

The platform reportedly only scans tracks upon upload and does not recheck outputs for potential infringement before export, which could facilitate the monetization of these AI-generated covers.

Uploading such content to streaming services via distribution platforms may allow for profit from copyrighted material without typical royalty payments to original composers.

Independent artists appear particularly vulnerable. Some original songs by artists on smaller labels or those self-distributing through platforms like Bandcamp or DistroKid have reportedly cleared Suno’s copyright filters without modification during testing, suggesting their material may be more susceptible to bypassing detection systems.

Reported Incidents

Several artists have reported incidents related to AI imitations appearing on streaming platforms:

Folk artist Murphy Campbell discovered AI covers of her YouTube songs uploaded to her Spotify profile. This led to false copyright claims against her original, public domain videos by a distributor, which were later rescinded following a social media campaign.

Experimental composer William Basinski and indie rock group King Gizzard and The Lizard Wizard have also experienced AI imitations of their work appearing on streaming platforms, sometimes diverting views from their official artist pages. Such incidents can particularly affect less famous musicians due to existing low payout thresholds on streaming services.

Industry Response

Suno has declined to provide comment on the observed filter bypasses.

Streaming services, including Deezer, Qobuz, and Spotify, have acknowledged the problem and stated they are implementing measures to address AI-generated spam and impersonations. Spotify has indicated its commitment to protecting artists' rights through safeguards, systems to identify duplicate or highly similar tracks, and human review processes. The company also notes the technical complexities involved with emerging technologies and the ongoing investment required for system evolution. The situation highlights broader systemic issues within music distribution and copyright enforcement.