Copyright law may be the key to protecting human creatives from AI

The biggest legal battle shaping the future of creative work may not be the one making headlines. While more than 90 lawsuits have been filed against AI companies for using copyrighted material to train their models, Jacob Noti-Victor and Xiyin Tang report for The Atlantic that a different copyright question will prove far more consequential: whether AI-generated works can receive copyright protection at all.

In 2024, a federal appeals court ruled in Thaler v. Perlmutter that works created autonomously by AI cannot be copyrighted, because copyright law requires a human author. The Supreme Court declined to review that ruling in March, leaving it in place.

The decision carries enormous economic weight. Entertainment and media companies operate by licensing intellectual property. Studios sell streaming rights, labels license recordings, and publishers sell adaptation rights across formats and languages. Copyright protection underpins every part of this system. Without it, anyone could freely copy or distribute a work, collapsing the entire financial model.

This creates a quiet but powerful incentive for major companies to keep humans involved in creative production. Netflix warns creators in its production guidelines against using AI to generate central characters or key visual elements without written approval. Publisher Hachette pulled a book after allegations emerged that parts of it were AI-generated. These decisions, the authors argue, are not acts of solidarity with human creators. They are acts of business pragmatism.

A recent example illustrates the stakes clearly. OpenAI announced a high-profile licensing deal with Disney to integrate beloved characters into its video-generation tool Sora. Within months, OpenAI shut Sora down entirely. Noti-Victor and Tang suggest the economics were simply unworkable: licensing content is expensive, and AI-generated material that cannot be copyrighted cannot become commercially viable intellectual property.

Not all creative sectors will benefit from this protection equally. Stock photography, for instance, is likely beyond saving. When companies can generate adequate images instantly, there is little reason to hire photographers or license from agencies like Getty. The argument holds most strongly for industries where large intermediaries still connect creators to audiences, including film, television, music, and book publishing. Audiences continue to value curation, and growing volumes of low-quality AI content appear to be strengthening that demand rather than weakening it.

The central challenge now is defining how much human involvement is required to make an AI-assisted work eligible for copyright. Industry players who stand to profit from replacing human labor will push for the loosest possible definition, where minimal prompting qualifies as authorship. The Copyright Office has indicated that prompting alone should not be sufficient, but courts have not yet confirmed this position.

Noti-Victor and Tang call for stricter penalties for misrepresenting AI involvement in copyright claims and urge courts and regulators to treat the copyrightability question as the central legal fight for human creative work, not a secondary concern.