Who Owns AI-Generated Content? Copyright & Ownership FAQ

Under current U.S. copyright law, AI-generated content without significant human creative input cannot be copyrighted. The U.S. Copyright Office has consistently held that copyright protection requires human authorship. In its March 2023 guidance, the Copyright Office stated that works created solely by AI are not eligible for copyright registration.

However, the ownership of AI outputs as a practical matter depends heavily on the Terms of Service of the AI platform you use:

• OpenAI (ChatGPT): Assigns all rights to outputs to users
• Anthropic (Claude): Similarly grants users rights to outputs
• Midjourney: Terms vary based on subscription tier, with paid subscribers generally owning their outputs

The key distinction is between legal copyright ownership (which may not exist for pure AI outputs) and contractual rights granted by the platform. Even without copyright, you may have contractual rights to use, sell, and commercialize AI-generated content under the platform's TOS.

The U.S. Copyright Office has issued clear guidance on AI-generated works through multiple rulings and policy statements. In February 2023, the Office ruled on the "Zarya of the Dawn" case, allowing copyright registration for human-authored elements (like story and arrangement) while denying protection for AI-generated images.

The March 2023 guidance document explicitly states that copyright requires human authorship, and works produced by machines without creative human involvement are not copyrightable. The Office distinguishes between works created "by" AI versus works created "with" AI.

If a human provides sufficient creative control, selection, and arrangement, the resulting work may qualify for copyright protection. The degree of human involvement is evaluated on a case-by-case basis. Factors include:

• The specificity of prompts
• The selection and arrangement of AI outputs
• Any human modifications to the final work

Simply typing a prompt is generally insufficient to establish authorship.

Yes, you can generally use AI-generated content commercially, but your rights depend on the specific platform's Terms of Service and applicable law.

• OpenAI: Terms of Service explicitly assign ownership of outputs to users and permit commercial use
• Anthropic Claude: Allows commercial use under their acceptable use policy
• Midjourney: Permits commercial use for paid subscribers, though free tier users have more limited rights

However, there are important caveats:

First, because pure AI outputs may not be copyrightable, competitors could potentially copy your AI-generated content without infringement.

Second, if the AI output infringes on existing copyrighted works (which can happen with image generators trained on copyrighted images), you could face liability.

Third, some industries have specific disclosure requirements for AI-generated content.

For maximum protection when commercializing AI content, add substantial human creative elements to establish copyrightable authorship.

The major AI platforms have different approaches to output ownership in their Terms of Service:

OpenAI (ChatGPT, DALL-E, GPT-4): Their terms assign all rights in outputs to users, subject to compliance with their usage policies. Users can use outputs for any purpose including commercial use. OpenAI retains no ownership claims to outputs.

Anthropic (Claude): Their terms grant users ownership of outputs generated through their services. Commercial use is permitted under their acceptable use policy. They focus more on use restrictions than ownership limitations.

Google (Gemini): Google's terms historically retained broader rights but have evolved. Current terms generally permit user ownership of outputs while Google retains rights to use inputs and outputs to improve services.

Microsoft (Copilot): As a ChatGPT-powered service, terms align with enterprise licensing. Business users typically own outputs under enterprise agreements.

All platforms reserve the right to use conversations to improve their models unless users opt out, and all have content policies that restrict certain types of outputs regardless of ownership.

The human authorship requirement is a foundational principle of U.S. copyright law that mandates creative works must originate from a human mind to receive copyright protection. This doctrine derives from the Copyright Act's reference to "authors" and has been reinforced by courts and the Copyright Office.

The Copyright Office's Compendium of Practices explicitly states that it "will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author."

For AI-generated content, this means:

• Purely AI-generated outputs with minimal human input are not copyrightable
• Works where humans provide detailed creative direction, selection, and arrangement may qualify
• The more human creative judgment involved, the stronger the copyright claim

To strengthen copyright claims for AI-assisted works:

• Use highly specific, creative prompts that reflect artistic choices
• Curate and select from multiple AI outputs
• Modify, arrange, and combine AI outputs with human creativity
• Document your creative process and decision-making

Even without copyright protection, several legal mechanisms can protect AI-generated content:

Contractual Rights: Platform Terms of Service grant you rights to use and commercialize outputs. You can include AI-generated content in contracts with clients and customers, creating enforceable obligations.

Trade Secret Protection: If you develop proprietary prompts or methodologies for generating AI content, these may qualify as trade secrets under the Defend Trade Secrets Act and state laws. Keep your processes confidential and use NDAs.

Trademark Protection: You can trademark brand names, logos, and distinctive elements associated with your AI-generated content, even if the underlying content isn't copyrightable.

Unfair Competition Laws: State unfair competition and misappropriation laws may provide recourse against competitors who copy your AI-generated content in bad faith.

Terms of Use: When publishing AI content, include Terms of Use that contractually restrict copying, even if copyright doesn't apply.

Compilation Copyright: If you select, arrange, and organize AI outputs in a creative way, the compilation itself may be copyrightable even if individual elements are not.

Disclosure requirements for AI-generated content depend on context, jurisdiction, and industry. Currently, there is no general federal law requiring AI disclosure in the United States, but this is evolving rapidly.

Specific Contexts Requiring Disclosure:

• Political advertising in some states requires AI disclosure
• The FTC considers non-disclosure of AI use potentially deceptive in certain commercial contexts
• Academic and journalistic standards typically require disclosure
• Professional services (legal, medical) may have ethical disclosure obligations

Platform Requirements:

• Many social media platforms require labeling AI-generated content
• Stock photo sites require AI disclosure for uploaded content
• Publishing platforms increasingly require AI contribution disclosure

Emerging Regulation: The EU AI Act requires disclosure for certain AI-generated content. Several U.S. states are considering or have passed AI disclosure laws. Federal legislation is under consideration.

AI systems trained on copyrighted works can sometimes generate outputs that substantially copy protected content, creating potential infringement liability.

Liability Framework:

• The user who publishes or distributes infringing AI output could be liable for direct infringement
• The AI company could face secondary liability claims for contributory or vicarious infringement
• Ongoing lawsuits (Getty Images v. Stability AI, New York Times v. OpenAI) are testing these theories

Platform Indemnification: Most AI platforms include indemnification provisions in their Terms of Service, but these vary significantly. OpenAI offers limited indemnification for enterprise customers. Coverage typically excludes intentional infringement or misuse.

Risk Mitigation:

• Use AI tools with content filters and safety measures
• Avoid prompts designed to replicate copyrighted works or styles
• Review AI outputs before publication for potential infringement
• Consider errors and omissions insurance that covers AI-related claims

Yes, you can register works that include AI-generated elements, but the Copyright Office requires disclosure and will only protect the human-authored portions.

Registration Requirements:

• You must disclose which portions of the work were AI-generated on the application
• The Office will limit registration to human-authored elements only
• Failing to disclose AI involvement could result in registration cancellation

Application Process:

• In the "Author Created" field, describe only the human-authored elements
• In the "Limitation of Claim" field, exclude AI-generated content from the claim
• Provide clear descriptions of what was human-created versus AI-generated

Strengthening Your Claim:

• Document your creative process and human contributions
• Retain evidence of significant modifications to AI outputs
• Show creative selection and arrangement of AI-generated elements

AI copyright law is rapidly evolving, with several potential changes on the horizon.

Legislative Developments: Congress is actively considering AI-related legislation, including disclosure requirements and liability frameworks. The EU AI Act establishes comprehensive regulations that may influence U.S. policy. Some proposals would create new IP categories specifically for AI-generated works.

Copyright Office Actions: The Office is conducting ongoing studies on AI and copyright, with public comments shaping future policy. New guidance on the human authorship requirement may provide more clarity. Registration procedures may evolve to better handle AI-assisted works.

Judicial Developments:

• Multiple pending lawsuits will clarify AI training data and output liability
• Courts will define the boundaries of fair use for AI training
• Decisions on AI authorship claims could reshape the human authorship doctrine

International Harmonization: Different countries are taking varied approaches, creating potential conflicts. International treaties may eventually address AI-generated works.