AI and U.S. Intellectual Property Law: Copyright, Patents, and Authorship

Artificial intelligence systems now generate text, images, music, code, and functional inventions at scale, forcing foundational questions about who — or what — can hold intellectual property rights under U.S. law. This page covers the intersection of AI with copyright doctrine, patent inventorship rules, and related authorship frameworks administered by the U.S. Copyright Office and the U.S. Patent and Trademark Office (USPTO). The stakes are significant: courts, federal agencies, and Congress are actively revising decades-old frameworks that were built around human creative and inventive agency.

• Definition and scope
• Core mechanics or structure
• Causal relationships or drivers
• Classification boundaries
• Tradeoffs and tensions
• Common misconceptions
• Checklist or steps (non-advisory)
• Reference table or matrix

Definition and scope

U.S. intellectual property law governing AI-generated works sits at the intersection of three doctrinal areas: copyright, patent, and trade secret law. Copyright protects original works of authorship fixed in a tangible medium (17 U.S.C. § 102). Patent law protects novel, non-obvious, and useful inventions as defined under 35 U.S.C. § 100 et seq.. Trade secret law, addressed separately in AI and Trade Secret Law, protects confidential commercial information.

The scope problem for AI is acute in two places. First, copyright requires human authorship — a doctrine the U.S. Copyright Office has enforced explicitly since its 1973 Compendium guidance and reaffirmed in its 2023 policy statement. Second, patent law requires a human "inventor" — a requirement the Federal Circuit confirmed in Thaler v. Vidal (43 F.4th 1207, Fed. Cir. 2022), holding that the plain language of 35 U.S.C. § 100(f) limits inventorship to natural persons.

These constraints shape the entire landscape of AI and intellectual property law, from how companies structure AI development workflows to how they document human contributions in patent applications.

Core mechanics or structure

Copyright: Human authorship doctrine

The U.S. Copyright Office registers works that reflect "original intellectual conception of the author" (Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 1884). The Office's Copyright and Artificial Intelligence Part 1 (2023) clarified that works generated entirely by AI without human creative control are not eligible for copyright registration. Works with sufficient human authorship — where a human selects, arranges, or modifies AI output — may qualify for registration covering only the human-authored elements.

The practical mechanics involve a sliding scale. A human who prompts an AI system and then selects, arranges, and edits outputs may retain copyright in the resulting selection and arrangement. The AI-generated elements themselves remain unprotected. The Copyright Office has applied this distinction in at least 3 published registration decisions since 2023, including the Zarya of the Dawn graphic novel case, where registration was narrowed to human-authored text while AI-generated images were excluded.

Patent: Inventorship and the natural person rule

Under 35 U.S.C. § 115, each patent application must name individuals who "invented or discovered" the claimed invention. The USPTO's February 2024 guidance (Federal Register Vol. 89, No. 37) confirmed that AI systems cannot be named as inventors. However, human inventors who use AI as a tool — and who themselves make a "significant contribution" to at least one claim — satisfy inventorship requirements.

The USPTO's "significant contribution" test draws from Pannu v. Iolab Corp. (155 F.3d 1344, Fed. Cir. 1998): each named inventor must contribute to the conception of the claimed invention. If an AI system independently conceives a novel solution and a human merely reduces it to practice without contributing to conception, that human may not qualify as inventor under current doctrine.

Causal relationships or drivers

Three structural forces drive the current unsettled state of AI IP law.

Technological capability outpacing statutory language. The Copyright Act of 1976 and the Patent Act of 1952 were drafted before generative AI existed as a commercial phenomenon. Neither statute addresses AI authorship or AI inventorship explicitly, leaving courts and agencies to apply language written for human actors.

The DABUS litigation series. Stephen Thaler's applications naming the AI system DABUS as sole inventor were rejected by the USPTO, the UK Intellectual Property Office, and the European Patent Office. The Federal Circuit's 2022 ruling in Thaler v. Vidal established binding U.S. precedent against AI inventorship. A parallel copyright case, Thaler v. Perlmutter (No. 22-1564, D.D.C. 2023), upheld the Copyright Office's refusal to register an entirely AI-generated image, affirming the human authorship requirement.

Commercial incentives and disclosure pressure. Companies deploying AI in R&D face structural incentives to document human contributions carefully. Failure to name correct inventors is grounds for patent invalidation under 35 U.S.C. § 256 and can constitute inequitable conduct. The AI patent inventorship question is therefore not merely academic — it carries direct validity risk for granted patents.

Congressional attention has increased. The AI in U.S. Legal System Overview context documents that multiple bills addressing AI IP have been introduced since 2022, though none had been enacted into statute as of the date of the USPTO's February 2024 guidance.

Classification boundaries

AI-related IP questions sort into four distinct categories with different legal treatment:

1. Purely AI-generated output — No human creative selection or inventive contribution. Not eligible for copyright or patent protection under current U.S. doctrine.

2. AI-assisted human creation — A human uses AI tools but exercises creative or inventive judgment over the final output or specific claims. Copyright and patent protection available for the human-contributed elements, subject to proof of contribution.

3. AI-generated output with human curation — A human selects, arranges, or edits AI-generated elements into a compilation. Copyright protects the selection and arrangement only, not the underlying AI-generated components (17 U.S.C. § 103).

4. Training data and model weights — Separate IP questions arise around the datasets used to train AI models. Reproducing copyrighted works in training datasets is the subject of active litigation (e.g., Andersen v. Stability AI, N.D. Cal., filed 2023). This area intersects with AI-generated content and copyright doctrine but is governed by fair use analysis under 17 U.S.C. § 107.

Tradeoffs and tensions

The human authorship and human inventorship requirements create a paradox: the more capable an AI system becomes, the harder it is to attribute protectable contribution to any specific human. This inverts the usual relationship between investment and protection — companies that invest the most in autonomous AI systems may receive the least IP protection for its outputs.

A second tension exists between disclosure and protection. Patent applications must disclose AI use under the USPTO's 2024 guidance, but disclosure of AI contribution risks undermining inventorship claims if human contribution cannot be independently demonstrated. This chilling effect on disclosure runs counter to the patent system's constitutional purpose of promoting public disclosure in exchange for limited monopoly (Article I, Section 8, Clause 8).

Copyright's human authorship rule also creates international divergence. The UK's Copyright, Designs and Patents Act 1988 (§ 9(3)) grants copyright in computer-generated works to "the person by whom the arrangements necessary for the creation of the work are undertaken" — a standard that would cover at least some AI outputs. This gap complicates cross-border IP strategy for companies operating in both jurisdictions.

A third tension involves the AI and trade secret law alternative: because copyright and patent protection may be unavailable, companies increasingly rely on trade secret law to protect AI models and outputs, which forecloses public disclosure entirely — the opposite of patent law's disclosure bargain.

Common misconceptions

Misconception: Prompting an AI system automatically creates copyright in the output.
Correction: The Copyright Office's 2023 guidance explicitly states that the "detailed prompts" a user provides do not transform AI-generated expression into human-authored work. Copyright requires that the human author be the one whose expression appears in the work — not merely the one who directed the machine.

Misconception: A company can hold copyright in AI-generated works through work-for-hire doctrine.
Correction: Work-for-hire under 17 U.S.C. § 101 requires a human employee or contractor. If no human author exists, there is no copyright to assign to an employer. The doctrine presupposes human authorship, not AI authorship.

Misconception: Listing a human as co-inventor alongside an AI system satisfies patent requirements.
Correction: The USPTO's 2024 guidance and Federal Circuit precedent hold that AI systems cannot be inventors. Listing an AI as co-inventor would render the application defective. The question is whether the human's contribution to conception is sufficient to qualify as the sole or joint inventor under the Pannu standard.

Misconception: Copyright in AI training data is irrelevant to the AI developer.
Correction: Multiple pending federal cases (including Andersen v. Stability AI and Getty Images v. Stability AI) allege that training on copyrighted images without license constitutes infringement. The outcome of these cases will directly shape the legal framework for AI model development.

Checklist or steps (non-advisory)

The following sequence describes the analytical steps that IP practitioners and organizations typically work through when assessing AI-related IP questions. This is a descriptive framework, not legal advice.

Step 1 — Identify the nature of human contribution
Document whether humans contributed to: (a) selection and arrangement of AI output (copyright); (b) conception of the specific claimed invention (patent); or (c) both.

Step 2 — Apply the Copyright Office human authorship test
Assess whether the human creative elements are sufficiently separable from AI-generated elements to support a narrowed registration claim, per the Copyright Office's 2023 guidance.

Step 3 — Apply the USPTO significant contribution test
For patent applications, evaluate each named inventor's contribution against the Pannu factors: (1) contribution to the conception of the claimed invention; (2) participation is not merely reducing to practice; (3) not merely following the directions of another.

Step 4 — Document AI tool use at time of creation
The USPTO's 2024 guidance requires applicants to disclose AI assistance that materially contributed to the invention. Contemporaneous records of human decision-making steps are critical for satisfying this requirement.

Step 5 — Assess trade secret as an alternative
If copyright and patent protection are unavailable or impractical, evaluate whether the AI model, training data, or outputs qualify for trade secret protection under the Defend Trade Secrets Act (18 U.S.C. § 1836).

Step 6 — Monitor international jurisdiction requirements
Apply jurisdiction-specific rules for any non-U.S. filings, as UK, EU, and Australian frameworks differ materially from U.S. doctrine on AI authorship and inventorship.

Step 7 — Track active litigation and regulatory guidance
The Copyright Office's AI study (ongoing as of Part 2, 2024) and USPTO rulemaking continue to evolve. Registration decisions and court rulings may shift applicable standards.

Reference table or matrix

References

• U.S. Copyright Office — Copyright and Artificial Intelligence Policy Statement (2023)
• U.S. Copyright Office — Title 17, U.S. Code
• USPTO — Inventorship Guidance for AI-Assisted Inventions, Federal Register Vol. 89, No. 37 (February 2024)
• 35 U.S.C. § 100 — Definitions (Patent Act)
• 35 U.S.C. § 115 — Oath or Declaration of Applicant
• 17 U.S.C. § 102 — Subject Matter of Copyright
• Defend Trade Secrets Act — 18 U.S.C. § 1836
• U.S. Constitution, Article I, Section 8, Clause 8 — Intellectual Property Clause
• USPTO — Patent Center and Filing Resources
• U.S. Copyright Office — Compendium of U.S. Copyright Office Practices, Third Edition