The Fight for 'Human-Made' Certification Is Becoming the Next Front in AI and Copyright Law

Murphy Campbell did not set out to become a test case for AI copyright law. The folk musician discovered AI-generated covers of her songs uploaded to Spotify under her name without her knowledge or consent — complete with AI-generated liner notes, fabricated biographical details, and listener reviews responding to music she had never made. By the time she identified the scope of what had happened, the AI-generated content had accumulated streaming revenue and audience that belonged, by any reasonable ethical standard, to her.

The Enforcement Gap

Campbell's experience illustrates a structural problem in the current copyright framework: the rules that protect against unauthorized reproduction of a specific creative work do not cleanly protect against the creation of convincing approximations of an artist's style and voice. An AI model trained on an artist's catalogue without licensing can generate new content that is stylistically indistinguishable from that artist's work — and existing copyright law has limited tools to address it, because the output isn't technically copying any specific protected expression.

The legal battles are beginning. The Recording Industry Association of America has filed suits against multiple AI music platforms for training data licensing. Several of those cases are likely to produce precedent-setting rulings on whether training constitutes infringement. But litigation is slow, jurisdiction-specific, and retrospective — it resolves past harms rather than preventing future ones.

The Certification Movement

What's gaining traction as a complementary approach is a push for a standardized "human-made" certification mark — a label system that would function like Fair Trade certifications for consumer goods, providing buyers, platforms, and listeners with a verified signal that content was produced without AI generation. Several organizations are developing competing frameworks, with the core design questions centering on how verification would work at scale, who would administer it, and what liability would attach to fraudulent certification.

The skeptical case is that such certifications would be unenforceable without platform cooperation — and platforms have strong economic incentives not to create friction around AI-generated content that performs well with audiences. The optimistic case is that cultural demand for authenticity creates a market signal that platforms will eventually have to respond to, particularly in segments where the "human-made" provenance is part of what audiences are paying for.

Campbell's case has become a galvanizing symbol for the certification movement precisely because it combines two distinct harms: economic displacement and identity fraud. Those are much harder to dismiss as abstract policy concerns than debates about training data licensing.