Suno's AI Music Tool Shows How Easily Copyright Filters Can Be Defeated

Suno, an AI platform for generating music, explicitly prohibits users from employing copyrighted material. Its system is designed to recognize and block attempts to replicate existing songs. However, new testing reveals these safeguards are surprisingly fragile, allowing users to create AI-generated tracks that closely mimic popular hits with simple, widely available workarounds.

Using Suno's Studio feature, which is part of a paid subscription, individuals can upload an audio file as a basis for new music. While the filter may catch an unaltered hit song, slowing a track down by half or doubling its speed in free software like Audacity often bypasses detection. Adding brief segments of white noise appears to guarantee success. After processing, the original tempo can be restored and the noise removed, leaving a copyrighted song as the core ingredient for AI generation.

The resulting outputs occupy a disquieting middle ground. They are clearly recognizable as derivatives of specific songs—from Beyoncé to Black Sabbath—yet feel hollow, lacking the dynamic nuance of a human performance. The instrumentals often flatten the original's artistic character into generic approximations.

This process violates Suno's own terms of service. More concerningly, the platform seems to scan only on upload, not re-checking final outputs for infringement. This creates a straightforward, if illicit, path for someone to export these AI covers and upload them to streaming services, potentially monetizing work that belongs to others.

Independent artists may be most at risk. During testing, songs by lesser-known musicians passed through Suno's copyright filter without any manipulation required. This echoes real-world incidents where AI-generated imitations of independent artists' work have appeared on major streaming platforms, sometimes diverting royalties.

While streaming services like Spotify state they employ safeguards and human review to combat unauthorized uploads, they acknowledge the ongoing challenge posed by evolving AI tools. Suno, for its part, has offered no public comment on these filter vulnerabilities, leaving a critical question unanswered in a system already straining under new technological pressures.

Source: The Verge