Detection results disclaimer

A result is an estimate

Every output of this tool is a probabilistic estimate derived from a small number of acoustic measurements, produced by the AI Music Detector analysis engine. It is not proof, not certification, not a forensic finding and not an expert opinion. It has not been validated against a labelled dataset — see accuracy.

Do not use a result to

• Accuse a person publicly of passing off generated music as their own
• Support a copyright claim, counter-claim or dispute
• Make an employment, commissioning, contractual or disciplinary decision
• Grade, penalise or investigate a student
• Remove, demonetise or ban content on a platform
• Substantiate a factual claim in journalism without independent corroboration

Reasonable uses

• Personal curiosity about an unfamiliar track
• A first, non-binding screening step before a human review
• One data point in an investigation that also has provenance evidence
• Teaching how synthetic-media detection works and where it fails
• Research on your own material

Both error types are real

Human music can be scored as AI-leaning, particularly when it is loudness-maximised or heavily compressed. Generated music can be scored as human-leaning, particularly when it has been re-mixed, re-recorded or produced by a recent model. Neither outcome is unusual.

No relationship with generators

This site is independent and not affiliated with, endorsed by or partnered with any AI music generation company. Product names are used only descriptively.

Formal terms

This disclaimer forms part of the terms of use, which contain the applicable warranty and liability provisions.