Key Takeaways
- Copyright law’s existing framework still governs AI-generated outputs: originality, human authorship and fixation remain the core principles.
- In 2025, the D.C. Circuit confirmed that the Copyright Act requires human authorship and does not permit copyright registration for works generated autonomously by AI.
- The Copyright Office has taken the same position: AI-assisted works may be protectable, but purely AI-generated material, or material reflecting insufficient human control over expressive elements, is not.
- For businesses, the practical question is not whether AI may be used to create content — it may. The question is whether the business can identify and document enough human-authored expression to support ownership, registration and enforcement.
Businesses across industries now use generative AI to draft advertising and website text, create images and presentations, generate software code and develop product concepts. As that use becomes more common, so does an increasingly important question: who, if anyone, owns the output?
The answer begins with familiar copyright doctrine. Section 102(a) protects only “original works of authorship,” and the Supreme Court has long held that originality requires independent creation plus at least a modicum of creativity.1 Copyright also depends on authorship, which the Court has described as the person “to whom anything owes its origin.”2 In the AI context, that makes human authorship the central issue.
Human Authorship Still Controls
The clearest recent appellate authority is Thaler v. Perlmutter. There, registration was sought for a visual work wholly generated by AI. The D.C. Circuit affirmed the refusal to register, holding that “the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being.”3 The court emphasized that authors are central to the statutory scheme and relied on Sections 102(a) and 201(a).
The Copyright Office’s 2023 registration guidance and 2025 AI report follow the same approach. The Office explained that when a machine produces a work’s “traditional elements of authorship,” the work lacks human authorship and is not registrable.4 The 2025 report likewise states that copyright does not extend to “purely AI-generated material, or material where there is insufficient human control over the expressive elements,” while recognizing that copyright may protect human-authored elements in AI-assisted works, including creative selection, arrangement, and modification.5
For businesses, then, the key inquiry is whether the final work contains enough human-authored expression to support copyright protection.
Prompting Alone Is Probably Not Enough
One of the clearest practical points in the Copyright Office’s guidance is that prompting, by itself, generally does not establish authorship. The 2025 report states that prompts alone do not provide sufficient control over the expressive elements of current AI outputs. The 2023 guidance likewise explains that prompts may describe the desired result, but the system determines how those instructions are expressed.
That does not mean prompts are irrelevant. They may help show process, and in some cases a prompt itself may be copyrightable as text if it is sufficiently original and fixed. But the copyright status of the prompt is separate from the copyright status of the output. Under the Copyright Office’s current view, the stronger ownership argument comes from meaningful human revision, selection, arrangement, transformation or incorporation of AI-generated material into a larger human-authored work.
Why Documentation Matters
This issue matters not only for registration, but also for enforcement and litigation. A timely registration remains important: under 17 U.S.C. § 410(c), a registration made before or within five years after first publication is prima facie evidence of validity and of the facts stated in the certificate.
That presumption, however, is rebuttable. In an AI-authorship dispute, a defendant may challenge validity through public statements, metadata, source files, version histories, prompt logs or other workflow evidence obtained in discovery. When validity is contested, the copyright owner may need to prove which portions of the work are actually human-authored and thus protectable.
Courts also do not treat a copyrighted work as entirely protected or entirely unprotected. The Copyright Office’s guidance contemplates mixed works and instructs applicants to claim the human-authored portions while excluding AI-generated material where appropriate. In practice, that means a court may protect only the human-created aspects of a work — such as substantial edits, creative arrangement or other original modifications — while treating the underlying AI-generated material as unprotectable.
That is why process matters. A company’s future position may depend less on broad claims that AI was merely a “tool” and more on whether it can show, with contemporaneous evidence, what a human actually contributed to the authorship of the copyrighted work.
Practical Steps Companies Should Consider Now
Businesses using generative AI should consider a few immediate steps:
- Document human creative input. Preserve drafts, edits, source files and records showing who selected, arranged, revised or transformed AI outputs.
- Do not assume prompting establishes ownership. Prompt logs may help show process, but they are not a substitute for evidence of human-authored expression.
- Be precise in registration strategy. Applicants should disclose AI-generated material where required, identify the human author’s contribution and claim only the human-authored portions.
- Revisit contracts and internal policies. Vendor agreements, employee policies and content-development protocols should not assume that all AI-assisted output is fully protectable or exclusively owned in the same way as traditionally authored material.
Bottom Line
The law remains grounded in conventional copyright principles. The Copyright Office’s recent guidance and the D.C. Circuit’s decision in Thaler point in the same direction: copyright protection still turns on human authorship, even when AI is part of the process.
For businesses, the practical implication is not simply that some AI-generated works may be unprotectable. It is that ownership, registration and enforcement may depend on whether the company can later show what a human author actually contributed. In that sense, AI governance is also evidence governance.
[1] 17 U.S.C. § 102(a); Feist, 499 U.S. at 345.
[2] Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 58 (1884).
[3] Thaler, 130 F.4th at 1041, 1044.
[4] 88 Fed. Reg. at 16,192–93.
[5] U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability iii, 15, 19–20.
