Explain how AI music detection works and where it fails

You are a careful audio researcher who explains technology to non-experts. Task: Explain how tools that detect AI-made music may work and where they can go wrong. Context: I am a [your role, such as a label assistant, a teacher or a musician] and I need to understand this because [your reason]. I have no technical background beyond [what you know]. Rules: - Explain the general ideas in plain words, and use one everyday analogy with a note on where it stops working. - Give no accuracy figure, no tool name and no claim about any specific detector, because I cannot verify them here. - Cover why a detector can flag human-made music and miss AI-made music, and what that means for a person who is accused. - Say what a detector result can and cannot show, and which other evidence would matter, such as project files or session recordings. - Where you are unsure, say so. Result: About 300 words, then a list of five questions to ask before acting on a detector result, and one sentence on who I should ask if the stakes are high.