Frame your creative relationship with AI
Priya is a senior graphic designer at a branding agency. A major client asks for a complete visual identity system. Under deadline pressure, Priya uses an AI design tool to generate 200 logo concepts based on a detailed brief she wrote. She narrows these to 15, then to 3, making modifications to each: adjusting proportions, refining typography, shifting color relationships. The client selects one. It becomes the face of a global campaign.
When the client's legal team draws up ownership documents, a dispute emerges. The AI tool's terms of service state that outputs are in the public domain. Priya's agency argues that her selection, refinement, and contextual judgment constitute authorship. The AI company maintains that the generative model performed the creative act. A competing designer, seeing the final logo, claims it closely resembles a concept she uploaded to the AI platform's community gallery two years ago.
The case forces everyone involved to reckon with what authorship actually means. Priya's brief was detailed and specific. Her selection process was informed by twenty years of design experience. Her refinements were substantial. But the underlying forms (the shapes, the spatial relationships, the typographic gestures) were generated by a model she did not build, trained on work she did not create.
Priya's agency hires an intellectual property lawyer who explains that current copyright law in most jurisdictions requires a human author. But the law has not caught up to the question of where the threshold of human contribution lies. "Writing a prompt is like writing a contract with a ghostwriter," the lawyer says. "The question is how much the ghostwriter contributed, and whether the ghostwriter is a person."
What do you think?
DISCUSSION QUESTIONS
• Is selecting from AI-generated options a creative act comparable to making something from scratch?
• Should the person who writes a detailed AI prompt be considered an author? What about someone who writes a vague one?
• If AI-generated outputs cannot be copyrighted, what protections should creators who refine them receive?
• How should credit be handled when multiple people's uploaded work influenced the AI's training data?
• Does the concept of sole authorship still make sense in an era of AI collaboration?
