Adoption as Authorship
- Author(s):
- Christopher Newman
- Posted:
- 09-2026
- Law & Economics #:
- 26-17
- Availability:
- Full text (most recent) on SSRN
ABSTRACT:
The U.S. Copyright Office treats expression generated by AI as categorically unprotectable, even when a human directs the process and incorporates the output into a work. This Article argues that the Office's position rests on a mistaken model of authorship, in which the author must personally generate each element of fixed expression or at least transcribe a precise mental conception formed in advance. Neither model describes how authors actually work. Creation proceeds by trial and error. Authors generate expressive variations, often through processes they do not control, and settle on results that fulfill an intent that may itself change along the way. Photographers shooting in burst mode, Ellsworth Kelly’s randomly dropped paper fragments, and Judge Frank’s “clap of thunder” in Alfred Bell & Co. v. Catalda all show that copyright already recognizes authorship of expression the author did not deliberately generate. The Article contends that the irreducible act of authorship is adoption, meaning the choice to treat a set of sensory signals as the embodiment of one's expressive intent. The joint authorship cases support this account. They locate authorship in decision-making authority over what the work will contain, not in the generation of content. The adoption theory does not license appropriating preexisting material, because originality doctrine separately bars claims to the commons or to others' expression. On this reading, originality works like the Lockean limits on appropriation. AI output is abundant and takes nothing from the commons, so there is no principled reason to deny protection to a human author who adopts it as her own expression.