AI Copyright in 2026: Who Actually Owns What AI Creates?
A marketing agency commissions a campaign built entirely on AI-generated images, copy, and video. The work is distinctive and commercially successful. Six months later, a competitor launches nearly identical creative — generated by the same AI model with similar prompts. The agency's lawyers discover that the agency may own no copyright in the campaign at all. The competitor copied material that was arguably never protected.
The question of who owns AI-generated work — or whether anyone does — is one of the most consequential open legal questions of the current technological era. The answers emerging from US courts, the Copyright Office, and international regulators in 2026 have direct practical implications for businesses building on AI-generated content.
Human Authorship Requirement — The principle, established by US courts and the Copyright Office, that copyright protection requires human creative expression. Works generated entirely by machines without human creative input cannot be copyrighted under current US law. The degree of human contribution required to establish copyright remains contested.
Thaler v. Vidal — A landmark US Federal Circuit Court case (2022) in which the court held that an AI system (DABUS) could not be listed as an inventor on a patent, because US law requires inventors to be natural persons. While technically a patent case, the reasoning has been influential in copyright discussions about AI authorship.
The US Position: Human Authorship Is Required
The US Copyright Office has issued formal guidance (2023, updated 2024) establishing that copyright protection requires human authorship. AI-generated content — content created entirely by an AI system without human creative contribution — cannot be registered and is not protected by copyright.
This position was crystallised in the Thaler v. Perlmutter case (DC District Court, 2023), where the court upheld the Copyright Office's refusal to register an AI-generated image. The court found that the "nexus between the human mind and creative expression" is an essential element of copyright, and that a work generated entirely by a machine without human creative input does not qualify.
The harder question is: what level of human contribution is sufficient? The Copyright Office has taken the position that human-curated selection and arrangement of AI-generated content can be protectable — if the selection reflects genuine human creative judgment. A human who prompts an AI to generate 100 images and selects 10 to form a coherent artistic collection may have copyright in the selection and arrangement, even if not in each individual image. A human who types a single generic prompt and accepts whatever the AI generates likely does not have a protectable creative contribution.
International Divergence: How Other Major Jurisdictions Approach AI Copyright
The US position — no human, no copyright — is not universal. Key international positions in 2026:
European Union: The EU has generally aligned with the human authorship requirement under the InfoSoc Directive and the emerging AI Act framework. The EU AI Act's transparency requirements for AI-generated content do not directly address copyright ownership, but the dominant view among EU member state courts is that purely AI-generated works lack copyright protection.
United Kingdom: The UK has a specific provision in the Copyright, Designs and Patents Act 1988 (Section 9(3)) that addresses "computer-generated works" — works with no human author — and assigns copyright to the person who makes the necessary arrangements for the work's creation. This potentially allows AI-generated works to receive copyright protection, attributed to the person who set up the AI system. This is an outlier position globally and is under active policy review.
China: Chinese courts have ruled in favour of copyright protection for AI-generated works in some cases, attributing copyright to the AI operator. A 2023 Beijing Internet Court ruling found that an AI-generated image could receive copyright protection attributable to the person who made creative choices in operating the AI. This is the most AI-permissive major jurisdiction position.
Japan: Japan's copyright agency has signalled that AI-generated content could receive protection if sufficient human creative contribution is involved, similar to the US approach but potentially more permissive in its assessment of what constitutes sufficient human input.
Practical Implications for Businesses Using AI-Generated Content
The copyright uncertainty around AI-generated content creates concrete business risks that organisations must manage:
No copyright protection means no exclusive rights. If your AI-generated content does not qualify for copyright protection (in the US, because no sufficient human creative contribution is present), any competitor can freely copy and use that content without infringement. This undermines the business rationale for investing in distinctive AI-generated brand assets.
Training data copyright risk runs the other direction. While AI outputs may not be copyrightable, the data used to train AI models may be copyrighted by its original creators. Major ongoing litigation — including lawsuits by news publishers, visual artists, and code authors against AI companies — alleges that training on copyrighted material without licence constitutes infringement. Businesses using AI models trained on potentially infringing data face potential downstream liability risk.
Contract clauses matter. When commissioning AI-generated creative work (from an agency or contractor), ensure your contract addresses: (1) who owns the AI-generated outputs; (2) what warranties the contractor gives about training data compliance; (3) who bears liability if a third party claims copyright infringement. Without explicit contractual allocation, these questions default to unclear legal positions.
What the Ongoing AI Copyright Cases May Resolve
Several cases working through the US federal courts in 2026 may clarify key outstanding questions. The Andersen v. Stability AI case (visual artists versus generative AI companies) and Getty Images v. Stability AI address whether training AI models on copyrighted works constitutes infringement. The outcomes will directly affect the training data risk exposure of all AI-generated content businesses.
Separately, the New York Times's lawsuit against OpenAI and Microsoft (alleging that training on and reproducing journalistic content infringes copyright) is the highest-profile case testing whether AI companies' use of published content for training can proceed without licence or compensation. If plaintiffs succeed on the key issues, it will significantly constrain the training data available to AI models and potentially require compensation frameworks for content creators whose work was used in training.
Frequently Asked Questions
Can I copyright AI-generated images or text?
In the US, not if the work was generated entirely by AI without sufficient human creative contribution. The Copyright Office requires human authorship. Works where a human makes genuine creative decisions about selection, modification, or arrangement of AI outputs may qualify for limited protection. In the UK, AI-generated works may receive copyright attributed to the person who arranged their creation under Section 9(3) CDPA. In China, some AI-generated works have received protection.
Does using AI-generated content expose my business to copyright infringement risk?
Potentially yes — from the training data direction. Major ongoing litigation alleges that AI models trained on copyrighted content infringe the rights of the original copyright owners. The outcomes of cases including Getty Images v. Stability AI and the New York Times v. OpenAI will affect this risk assessment. Review your AI vendor's terms regarding training data and obtain appropriate representations.
Who owns AI-generated content commissioned from an agency?
Whoever the contract says. Without explicit contractual allocation, ownership is legally uncertain. Commission contracts should specify: who owns the AI outputs, what warranties the agency provides about training data compliance, and who bears liability for third-party copyright claims. Do not assume the commissioning party automatically owns AI-generated work-for-hire.
How can a business establish copyright in AI-assisted creative work?
Ensure and document genuine human creative contribution: humans who make substantive decisions about what to keep, modify, discard, arrange, or enhance in the AI's output. A creative workflow that treats AI as a drafting tool — with humans making final creative choices — is more defensible than one where the AI output is approved wholesale. Document the human creative contributions in your production records.
This article is published by an independent news publication for informational purposes only and does not represent or claim affiliation with any government body, international organization, or official authority.