Humans get sued for quoting lyrics while AI gets a pass, sparking calls for new laws

If a human writer wants to quote a 12-word song lyric in a 75,000-word memoir, publishers will sound the alarm over ruinous licensing fees. But if a trillion-dollar tech giant scrapes every lyric ever written to power a commercial chatbot, U.S. courts seem ready to call it fair use.

That paradox is at the heart of a recent piece by Guardian columnist Alexander Hurst. While working on a book about trading his way up to $1.2 million and crashing back down to zero, Hurst wanted to use a single line from the National: “I still owe money to the money to the money I owe.”

His publisher gave him a flat no. Lyric rights are famously guarded, and buying clearance is often astronomical. Quoting Eminem's Lose Yourself was equally off the table.

The double standard in fair use

Yet, as Hurst pointed out, asking ChatGPT or Claude to write an Eminem-style rap incorporating that exact National lyric generates an instant reproduction without a dime going to the songwriters.

Following recent court rulings in lawsuits brought by authors against Meta and Anthropic, AI companies have gained significant ground under fair use arguments. Judges have largely treated ingestion and statistical word correlation as transformative training rather than direct copying.

Music publishers are not backing down quietly. Eight Mile Style, the company behind Eminem's catalog, has filed its own lawsuit against Meta, accusing the tech giant of exploiting artists' work for obscene corporate profits without securing licenses.

For LLMs, the output is the product. While legal scholars argue that statistical learning resembles note-taking, artists see a wholesale heist of human culture disguised as computer science.

Why it matters

We think this dispute exposes the growing absurdity at the intersection of intellectual property and AI.

For everyday writers, musicians, and independent creators, the playing field looks completely rigged. A solo author can be sued into oblivion for referencing a dozen words in print, while Big Tech deploys armies of attorneys to shield commercial models trained on billions of protected lines. Creators simply cannot afford to test fair use in court, leaving them exposed while tech platforms monetize their work.

For the AI industry, these early courtroom wins may feel reassuring, but they risk accelerating a severe regulatory backlash. When laws produce technically defensible outcomes that feel intuitively unfair, pressure mounts to rewrite the statutes entirely.

If AI is going to disrupt every aspect of modern society, the century-old framework of copyright law is likely next on the chopping block.