{"id":816,"date":"2026-05-28T02:37:14","date_gmt":"2026-05-28T02:37:14","guid":{"rendered":"https:\/\/www.perspireip.com\/blog\/?p=816"},"modified":"2026-05-28T02:37:14","modified_gmt":"2026-05-28T02:37:14","slug":"ai-copyright-law-design-ownership","status":"publish","type":"post","link":"https:\/\/www.perspireip.com\/blog\/ai-copyright-law-design-ownership\/","title":{"rendered":"AI and Copyright: Who Owns AI-Generated Designs in 2026?"},"content":{"rendered":"\n<p>Every week someone asks us the same question: &#8220;I used an AI tool to design my logo \/ draft my campaign artwork \/ generate a product render \u2014 do I own it, can I copyright it, and is there a law against this?&#8221; The honest answer in 2026 is that <strong>AI copyright law<\/strong> exists, it is moving fast, and the rules are not where most marketers and product teams assume they are. The U.S. Copyright Office has issued formal guidance. The EU AI Act has hard obligations now in force. And the courts have already drawn a line that disqualifies pure prompt-and-go AI outputs from any kind of registered protection.<\/p>\n\n\n\n<p>This article walks through what AI copyright law actually says about AI-generated works and designs, how ownership works in practice, what controlling measures companies are putting in place, and which AI-specific laws you need to know in the U.S., EU, and elsewhere. We are an IP services firm, not an AI vendor, so we will tell you where the actual legal traps are rather than where the marketing decks claim everything is fine.<\/p>\n\n\n\n<div class=\"wp-block-rank-math-toc-block\" id=\"rank-math-toc\"><h2>Table of Contents<\/h2><nav><ul><li><a href=\"#what-ai-copyright-law-says\">What AI Copyright Law Actually Says About AI-Generated Works<\/a><\/li><li><a href=\"#who-owns-it\">Who Owns an AI-Generated Design \u2014 and Who Doesn&#8217;t<\/a><\/li><li><a href=\"#controlling-measures\">How Companies Are Setting Controlling Measures for AI Outputs<\/a><\/li><li><a href=\"#ai-laws\">Is There a Law Specifically About AI? Yes \u2014 Several<\/a><\/li><li><a href=\"#real-world-examples\">Real-World Examples of AI Copyright Law in Action<\/a><\/li><li><a href=\"#how-perspireip-can-help\">How PerspireIP Helps Brands Navigate AI Copyright Law<\/a><\/li><li><a href=\"#conclusion\">Conclusion<\/a><\/li><li><a href=\"#faq\">Frequently Asked Questions<\/a><ul><li><a href=\"#can-i-copyright-something-i-made-with-an-ai-tool\">Can I copyright something I made with an AI tool?<\/a><\/li><li><a href=\"#who-owns-an-ai-generated-design-the-user-the-ai-company-or-no-one\">Who owns an AI-generated design \u2014 the user, the AI company, or no one?<\/a><\/li><li><a href=\"#is-there-a-law-specifically-about-ai\">Is there a law specifically about AI?<\/a><\/li><li><a href=\"#can-an-ai-system-be-named-as-an-inventor-on-a-patent\">Can an AI system be named as an inventor on a patent?<\/a><\/li><li><a href=\"#what-controlling-measures-should-companies-put-in-place-for-ai-generated-content\">What controlling measures should companies put in place for AI-generated content?<\/a><\/li><\/ul><\/li><\/ul><\/nav><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/images.unsplash.com\/photo-1620712943543-bcc4688e7485?w=1600&amp;q=80&amp;auto=format&amp;fit=crop\" alt=\"AI copyright law affecting AI-generated designs and machine learning models\"\/><figcaption class=\"wp-element-caption\">Who owns the output of an AI model? In 2026, the law has answers \u2014 but they are not the ones most users expect.<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"what-ai-copyright-law-says\">What AI Copyright Law Actually Says About AI-Generated Works<\/h2>\n\n\n\n<p>Start with the rule that controls everything else: in the United States, copyright protects only works of <em>human<\/em> authorship. That principle is older than AI; it is what disqualified the famous &#8220;monkey selfie&#8221; from registration in 2018. In 2025 the U.S. Copyright Office formalized how the rule applies to generative AI in <a href=\"https:\/\/www.copyright.gov\/ai\/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf\" target=\"_blank\" rel=\"noopener\">Part 2 of its Copyright and Artificial Intelligence report<\/a>. Two sentences from that report do most of the work.<\/p>\n\n\n\n<p>First: a work that is &#8220;entirely generated by AI&#8221; is not copyrightable, full stop. Second: a work that includes AI-generated material can still be registered, but only as to the human contributions \u2014 and those contributions have to amount to actual creative expression, not just typing a prompt. Prompts alone, no matter how clever, do not count, because the AI model decides which of millions of possible outputs to generate. The applicant must disclose AI-generated portions and limit the claim to what the human actually authored.<\/p>\n\n\n\n<p>Designs (in the U.S. design-patent sense) face a parallel rule under patent law. In the DABUS case, the Federal Circuit held that an AI system cannot be named as an inventor on a U.S. patent \u2014 only a natural person qualifies. The Supreme Court declined to revisit the issue in 2026. The USPTO&#8217;s 2024 AI inventorship guidance allows AI-assisted inventions and AI-assisted designs to be patented, but every claim must trace back to a significant human contribution. Push the AI button and accept the first render unchanged, and you have nothing to register.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"who-owns-it\">Who Owns an AI-Generated Design \u2014 and Who Doesn&#8217;t<\/h2>\n\n\n\n<p>Ownership of AI output is mostly a matter of three overlapping layers: the law (covered above), the terms of service of the AI tool you used, and any contracts between the user and a downstream client. The default under U.S. AI copyright law is that no one holds copyright in a purely AI-generated image, song, or text \u2014 it falls into a strange kind of unprotected zone where anyone can copy it, but no one can stop them, including the person who paid to generate it.<\/p>\n\n\n\n<p>The terms of service of major AI tools layer on top of that. OpenAI, Midjourney, Adobe Firefly, and similar platforms typically assign whatever rights exist in the output to the user \u2014 but they cannot grant copyright where copyright does not exist in the first place. They can only grant a contractual license over the data file. That is enough for most commercial uses (you can put the image on your website), but it is not enough to sue a competitor who copies the same image and uses it on theirs.<\/p>\n\n\n\n<p>Trademarks are different. AI-generated logos can be registered as trademarks because trademark protection is about source identification, not authorship. If your AI-generated logo is used in commerce to identify your goods, you can file at the USPTO and obtain a registration the same way you would for a human-designed logo. The &#8220;human authorship&#8221; rule simply does not apply to marks. For a refresher on trademark filing, see our <a href=\"https:\/\/www.perspireip.com\/blog\/trademark-clearance-search-guide\/\">trademark clearance search guide<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"controlling-measures\">How Companies Are Setting Controlling Measures for AI Outputs<\/h2>\n\n\n\n<p>Once you accept that AI copyright law leaves most pure-prompt outputs unprotected, the practical question becomes: what controls do you put in place so your team is not building a brand on unprotectable assets? Five measures have become standard at companies we work with.<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>Mandatory human editing.<\/strong> Anything generated by AI goes through a documented human-edit pass before it gets used in marketing or product. The edits are saved as versioned files so you can later demonstrate the human creative input if you need to register.<\/li>\n\n\n\n<li><strong>AI disclosure log.<\/strong> Every AI-touched asset is logged with the model, prompt, date, and reviewer. If a registration question arises, you can answer the Copyright Office&#8217;s mandatory disclosure honestly.<\/li>\n\n\n\n<li><strong>Training-data risk review.<\/strong> Before relying on a model&#8217;s output for a customer-facing campaign, IP counsel reviews the model&#8217;s training-data disclosures. Models trained on opted-out or unlicensed content carry a higher infringement risk for the outputs.<\/li>\n\n\n\n<li><strong>Tool-tier policies.<\/strong> Free-tier consumer AI tools usually have weaker indemnities than enterprise tiers. Most large companies have shifted to enterprise contracts that include IP indemnification \u2014 Microsoft, Adobe, and Anthropic offer them.<\/li>\n\n\n\n<li><strong>Trademark-first protection.<\/strong> Where a brand cannot rely on copyright, the next-best protection is trademark. Names, taglines, and logos can be registered even if they were AI-generated, and registration creates real enforceable rights.<\/li>\n<\/ol>\n\n\n\n<p>For broader context on how IP fits into a deal-ready portfolio, our piece on <a href=\"https:\/\/www.perspireip.com\/blog\/ip-due-diligence-business-deal\/\">IP due diligence for business deals<\/a> covers the questions investors and acquirers now ask about AI-generated assets specifically.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"ai-laws\">Is There a Law Specifically About AI? Yes \u2014 Several<\/h2>\n\n\n\n<p>This is the question that gets asked most often, and the answer in 2026 is that the law is now substantial and getting denser. The headline pieces are the EU AI Act, the U.S. Copyright Office&#8217;s AI reports, the USPTO inventorship guidance, and a growing wave of training-data lawsuits in U.S. federal court.<\/p>\n\n\n\n<p>The <a href=\"https:\/\/digital-strategy.ec.europa.eu\/en\/policies\/regulatory-framework-ai\" target=\"_blank\" rel=\"noopener\">EU AI Act<\/a> entered key effective dates in 2025 and 2026. General-purpose AI model providers must now publish a sufficiently detailed summary of the content used to train their models, including data protected by copyright, and they must put in place a copyright policy that respects opt-outs under the EU Copyright Directive&#8217;s text and data mining exception. The European Commission&#8217;s mandatory template for that public training-data disclosure took effect on August 2, 2025. Models placed on the market before that date have until August 2, 2027, to publish.<\/p>\n\n\n\n<p>In the United States there is no single federal AI statute, but there is a thicket of related rules: the Copyright Office&#8217;s human-authorship guidance, the USPTO&#8217;s AI inventorship guidance, state laws on AI-generated content disclosures (California, Tennessee, Colorado all passed measures between 2024 and 2025), and an active stream of copyright litigation \u2014 most prominently <em>The New York Times v. OpenAI<\/em>, <em>Getty Images v. Stability AI<\/em>, and a wave of class actions from artists and authors. None of these has produced a final, settled answer yet, but rulings in 2025 narrowed certain fair-use defenses and increased the legal exposure for AI vendors that train on unlicensed copyrighted material.<\/p>\n\n\n\n<p>Other jurisdictions have moved more cautiously. The UK has held that AI cannot be an inventor (the Supreme Court confirmed in <em>Thaler v. Comptroller<\/em>), Japan allows broad AI training under its copyright exception, and China has registered some AI-generated images but requires demonstrable human creativity. The net effect: the regime you operate under depends on where you generate, where you publish, and where you sell.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"real-world-examples\">Real-World Examples of AI Copyright Law in Action<\/h2>\n\n\n\n<p>A consumer-goods client came to us last quarter wanting to copyright a packaging illustration generated by Midjourney. The honest advice was no \u2014 the illustration was pure prompt output, the Copyright Office would reject the claim, and any registration that did slip through would be vulnerable to cancellation. We re-scoped the work: a human illustrator used the AI output as a reference, redrew significant elements, and the final composition was registered as a derivative work with the AI-generated portions disclosed and excluded from the claim. The illustrator owned the new layer of expression, and the company licensed it.<\/p>\n\n\n\n<p>Another example: a B2B SaaS client wanted to protect a UI design that had been roughed in by an AI design tool and then heavily refined by an in-house product designer over three weeks. That refinement was the registrable creative input. The application was filed with a clear claim limited to the human design contribution. It registered without objection. The takeaway from both cases is the same: AI-touched work is not automatically unprotectable \u2014 it is unprotectable only when there is no real human creative layer on top.<\/p>\n\n\n\n<p>A third case worth noting on the training-data side: a media company sued an AI vendor whose model had regurgitated near-verbatim copies of the company&#8217;s articles. Discovery showed the model had been trained on those articles without a license. The case is ongoing but has already produced settlements with other plaintiffs in the same range as mid-tier patent litigation. For users of AI tools, the practical effect is that enterprise indemnification clauses now matter \u2014 using a vendor without one transfers the risk to your company.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"how-perspireip-can-help\">How PerspireIP Helps Brands Navigate AI Copyright Law<\/h2>\n\n\n\n<p>PerspireIP&#8217;s AI-IP practice does three things. We audit your current AI workflows and tell you which outputs are protectable, which need restructuring, and which carry training-data infringement risk. We file copyright and trademark registrations on AI-assisted works with the human-authorship disclosures the Copyright Office now requires, structured to maximize the chance of a clean registration. And we draft AI-use policies and vendor contracts so your team is using enterprise-tier tools with IP indemnification, not free-tier tools that leave you exposed. The result is a workflow that uses AI for the speed and cost benefits without trading away the IP that ought to be defending your brand.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"conclusion\">Conclusion<\/h2>\n\n\n\n<p>AI copyright law in 2026 is no longer a frontier \u2014 it is a defined legal terrain. Pure AI-generated works are not copyrightable. AI-assisted works are, if there is real human creative input that is properly disclosed. AI cannot be named as an inventor on a patent. The EU AI Act now requires training-data transparency and copyright policy compliance for general-purpose models. And the U.S. courts are working through high-stakes training-data lawsuits that will reshape vendor indemnification for years to come. The companies that win in this environment will not be the ones that move fastest or slowest with AI; they will be the ones that put the right human creative layer, the right disclosures, and the right vendor contracts in place. If you want a clear-eyed read on where your AI workflow sits today, <a href=\"https:\/\/www.perspireip.com\/contact\/\">talk to PerspireIP<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"faq\">Frequently Asked Questions<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"can-i-copyright-something-i-made-with-an-ai-tool\">Can I copyright something I made with an AI tool?<\/h3>\n\n\n\n<p>Only the human-authored portions. If you used a prompt and accepted the output unchanged, the U.S. Copyright Office will not register it. If you made meaningful creative edits \u2014 composing, redrawing, arranging, modifying \u2014 those edits are registrable, and the AI-generated portions must be disclosed and excluded from the claim.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"who-owns-an-ai-generated-design-the-user-the-ai-company-or-no-one\">Who owns an AI-generated design \u2014 the user, the AI company, or no one?<\/h3>\n\n\n\n<p>Under U.S. AI copyright law, no one holds a copyright in a purely AI-generated design because there is no human author. The AI company&#8217;s terms of service usually assign any rights that do exist in the output to the user, but those rights are contractual licenses to the data file, not copyright ownership. Trademark protection is still available for AI-generated logos used in commerce.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"is-there-a-law-specifically-about-ai\">Is there a law specifically about AI?<\/h3>\n\n\n\n<p>Yes. The EU AI Act took effect in 2024 with key transparency and copyright obligations for general-purpose AI providers kicking in August 2025. In the U.S., the Copyright Office and USPTO have issued formal guidance documents, and several states (California, Tennessee, Colorado) have AI-specific disclosure laws. There is also a growing body of federal court litigation on AI training data.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"can-an-ai-system-be-named-as-an-inventor-on-a-patent\">Can an AI system be named as an inventor on a patent?<\/h3>\n\n\n\n<p>No. The U.S. Federal Circuit ruled in Thaler v. Vidal that only a natural person can be named as an inventor. The Supreme Court declined to take the case in 2026. AI-assisted inventions are patentable as long as a natural person made a significant contribution to each claim.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"what-controlling-measures-should-companies-put-in-place-for-ai-generated-content\">What controlling measures should companies put in place for AI-generated content?<\/h3>\n\n\n\n<p>Five practical measures: a mandatory human-edit pass on AI output, an AI disclosure log per asset, a training-data risk review for high-stakes uses, enterprise-tier AI tool contracts with IP indemnification, and a trademark-first protection strategy for assets that copyright cannot cover.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>AI copyright law in 2026: who owns AI-generated designs, what the USPTO and EU AI Act actually say, and how brands should respond.<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[175,176,178,180,179,177,99,80],"class_list":["post-816","post","type-post","status-publish","format-standard","hentry","category-ip-strategy","tag-ai-copyright-law","tag-ai-generated-design","tag-ai-inventor","tag-ai-policy","tag-copyright-office","tag-eu-ai-act","tag-trademark","tag-uspto"],"_links":{"self":[{"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/posts\/816","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/comments?post=816"}],"version-history":[{"count":2,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/posts\/816\/revisions"}],"predecessor-version":[{"id":818,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/posts\/816\/revisions\/818"}],"wp:attachment":[{"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/media?parent=816"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/categories?post=816"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.perspireip.com\/blog\/wp-json\/wp\/v2\/tags?post=816"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}