60. AI and Copyright (Beginner Guide): What Creators Should Know About Using AI‑Generated Text and Images

Last Updated: June 5, 2026

AI and copyright is no longer a theoretical debate for legal scholars — it is an active battleground with real court rulings, record-breaking settlements, and new guidance from regulators on both sides of the Atlantic. In 2026, the question of whether AI-generated content can be copyrighted has a clearer answer than ever before, even as the larger question of whether AI companies can legally train on copyrighted works without permission remains actively contested in courtrooms across the US and Europe. If you are using AI to write, design, build products, or create content for clients, understanding where the law currently stands is not optional.

This guide covers everything creators and businesses need to know about AI and copyright in 2026. You will learn what the U.S. Copyright Office’s landmark three-part AI report concluded, which court cases are setting precedent right now, how the rules differ between the United States, the European Union, and the United Kingdom, and what practical steps you can take today to protect your work and reduce your legal exposure. The article also covers commercial AI image tool licensing, indemnification policies, and what businesses using AI content must know before publishing or selling AI-assisted work.

Whether you are a freelance creator using AI tools daily, a marketing team publishing AI-assisted content at scale, or a business leader evaluating your organization’s AI content policy, the 2026 copyright landscape has specific implications for you. Courts are splitting on fair use for AI training data. The Supreme Court has cemented the human authorship requirement. And major AI companies have already paid billions to settle copyright claims. The rules are being written in real time — and this guide keeps you current as of June 2026.

⚖️ 1. Can AI-Generated Content Be Copyrighted? The 2026 Answer

The short answer is: it depends on how much of a human created it. Purely AI-generated content — text written entirely by a chatbot from a prompt, or an image produced by typing a description into an image generator — cannot be copyrighted under US law as it stands in 2026. This is not a gray area. The U.S. Copyright Office has stated it clearly, the courts have affirmed it, and the Supreme Court declined to revisit the issue in March 2026, leaving the rule intact.

The governing principle is the human authorship requirement, which is rooted in the U.S. Constitution’s Copyright Clause. The Copyright Office’s Part 2 report, published January 29, 2025, reaffirmed this position in plain terms. The report concluded that “given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output.” Entering a detailed prompt into ChatGPT or Midjourney, no matter how specific or creative the wording, does not make you the legal author of what comes out. The AI system — not you — determines the expressive elements.

However, AI-assisted content is a different matter. When a human author uses AI as a tool but retains meaningful creative control over the final work — selecting, arranging, editing, or substantially modifying the AI’s output — that human contribution can attract copyright protection. The Copyright Office has registered hundreds of works containing AI-generated material where the human contribution was sufficient. The line is not about whether AI was involved; it is about whether a human being made the expressive choices that define the final work. For practical purposes, this means: the more you edit, arrange, and creatively transform AI output, the stronger your copyright claim. The more you copy and paste without modification, the weaker — or nonexistent — your protection.

The 2026 Copyright Reality: Purely AI-generated content cannot be copyrighted in the US, EU, or UK. AI-assisted content, where a human author exercises meaningful creative control, can be. The threshold question is always: who made the expressive choices?

⚖️ 2. Key AI Copyright Court Cases in 2025–2026

The most consequential developments in AI copyright law in the past eighteen months have not come from legislatures — they have come from courts. As of June 2026, more than 160 active copyright lawsuits are filed against AI companies in the United States alone, covering everything from news articles and books to song lyrics, photographs, and software code. A handful of landmark cases and rulings have begun to define the legal landscape, though the most important question — whether training AI models on copyrighted works constitutes fair use — remains unresolved and is unlikely to receive a definitive appellate ruling before late 2026 at the earliest.

Thaler v. Perlmutter — Human Authorship Cemented. The most settled area of AI copyright law concerns who can hold copyright — and the answer, definitively, is: only a human being. In March 2025, the U.S. Court of Appeals for the D.C. Circuit affirmed that an AI system cannot be an author under the Copyright Act. The Supreme Court denied certiorari in this case on March 2, 2026, meaning the D.C. Circuit’s ruling stands and no other appellate court is positioned to reopen the question. For creators and businesses, this matters because it closes the door entirely on registering AI-generated works without human creative contribution. It does not, however, prevent copyright protection for works where humans made the expressive choices.

New York Times v. OpenAI & Microsoft — Training Data on Trial. The New York Times sued OpenAI and Microsoft in December 2023, alleging that millions of its articles were copied without permission to train ChatGPT. The Times is seeking “billions” in statutory damages and claims that ChatGPT outputs can reproduce its paywalled content, directly competing with its business. As of June 2026, the case is deep in discovery — a court ordered OpenAI to produce tens of millions of output logs — and no trial date has been set. This case is one of the most closely watched in the industry because a ruling for the Times could force AI companies to pay licensing fees for training data across the media industry, fundamentally altering the economics of AI development.

Bartz v. Anthropic — The Largest Copyright Settlement in US History. Three authors sued Anthropic for training Claude on pirated books. The case resulted in a $1.5 billion settlement — the largest copyright settlement in U.S. history — with a final fairness hearing held in April 2026. Separately, a court found that Anthropic was liable for maintaining a “central library” of pirated books not directly connected to training. This dual finding — one count settled, another creating additional exposure — illustrates that AI companies face copyright risk not only for what their models learn, but for how they store and organize training material.

GEMA v. OpenAI — Europe’s First Major Ruling Against an AI Company. In November 2025, the Munich Regional Court (Landgericht München I, Case No. 42 O 14139/24) held that OpenAI’s use of copyrighted German song lyrics to train its GPT models violates German copyright law. This was among the first major European AI training rulings against an AI developer and signals that European courts may take a stricter approach to training data than their US counterparts. The case may eventually reach the Court of Justice of the European Union, which could produce the first CJEU ruling on AI and copyright.

Getty Images v. Stability AI — UK High Court Delivers a Mixed Result. In November 2025, the UK High Court delivered its judgment in Getty Images v. Stability AI ([2025] EWHC 2863 (Ch)). The court rejected Getty’s claim that Stable Diffusion model weights are “infringing copies” under UK copyright law. However, Getty won on limited trademark claims related to the reproduction of its watermarks in AI outputs. The case established that model weights themselves may not constitute direct copyright infringement — but that outputs reproducing identifiable elements of copyrighted works, such as watermarks, can still trigger trademark liability. An appeal on related jurisdiction issues is expected before December 2026.

The Judicial Divide on Fair Use. As of June 2026, three US district judges have ruled on whether AI training on copyrighted works constitutes fair use — and they have reached different conclusions. One judge found training to be transformative and dismissed market harm concerns. A second sided with the AI company but cautioned that AI training “in many circumstances” would not qualify as fair use, raising concerns that generative AI could flood the market with content and erode incentives for human creators. A third found against the AI company on the market harm factor. This three-way split means the question will almost certainly need appellate resolution — likely in the Third or Ninth Circuit — before any clear national standard emerges. The Copyright Office’s Part 3 report, released in pre-publication form in May 2025, concluded that “some uses of copyrighted works for generative AI training will qualify as fair use, and some will not” — a position that reflects the judicial uncertainty exactly.

🌍 3. AI Copyright by Jurisdiction — US vs EU vs UK

One of the most practically important things for creators and businesses to understand in 2026 is that AI copyright law is not uniform across jurisdictions. The rules in the United States, the European Union, and the United Kingdom differ significantly on training data, ownership of AI-assisted works, and disclosure requirements. Organizations operating across borders — or publishing content for international audiences — need to understand which rules apply to them. What is legally defensible in one jurisdiction may constitute infringement in another.

The EU has taken a structured approach through the Digital Single Market (DSM) Directive, which created a general text and data mining (TDM) exception allowing AI training on lawfully accessible content — but with a critical opt-out mechanism. Rights holders who expressly reserve their rights can prevent their works from being used for training. The EU AI Act added further requirements: general-purpose AI model providers must publish a summary of their training data and implement a policy to comply with opt-outs. However, significant uncertainty remains about how these opt-outs work in practice and whether they are being respected, with many EU member states questioning the enforceability of the mechanism.

The UK took a notably different turn in March 2026. After a lengthy consultation process that received over 11,500 responses, the UK government confirmed it will not introduce a broad copyright exception for AI training. Instead, the existing framework applies: AI developers must obtain licences unless a specific existing exception — such as the non-commercial research exception under Section 29A of the Copyright, Designs and Patents Act 1988 — covers the use. This places the UK among the jurisdictions with greater protection for rights holders, though it also creates uncertainty for AI developers operating in the UK market. The UK is monitoring global legal developments and has explicitly stated it will not rush legislative reform.

Note: Legal positions accurate as of June 2026. Consult a qualified IP attorney for advice specific to your situation and jurisdiction.

✅ 4. Safe AI Content Use — A Practical Framework for Creators

Understanding what the law says is only half the job. The more important question for most creators is: what should I actually do to protect my work and avoid liability? The 2026 consensus from the Copyright Office, legal practitioners, and AI tool providers points to a clear set of practices that significantly reduce risk — for both your intellectual property and your clients’. The framework below is organized by content type and applies whether you are a solo creator or a team publishing AI-assisted content at scale.

For Text Content

The safest way to use AI for text in a commercial context is to treat it as a powerful drafting and editing assistant — not a sole author. When you use an AI tool to produce a first draft and then substantially edit, restructure, add your own analysis, and apply your own creative judgment to the final piece, you are making the expressive choices that the Copyright Office requires for human authorship. The more transformation you apply, the stronger your copyright claim. Read our guide on building a safe AI content publishing workflow for a step-by-step team process.

• Use AI as a drafting and editing tool — not as a sole author or final publisher
• Always add substantial human creative input: analysis, restructuring, original examples, your voice
• Never reproduce copyrighted text verbatim in your prompts — this can cause the AI to output infringing content
• Review all AI output carefully for potential reproduction of copyrighted material before publishing
• Document your human contributions — keep drafts, notes, and edit histories that demonstrate your creative process
• Disclose AI use where required by platform terms (Google, LinkedIn, Amazon all have specific policies)

For Images

AI image generation carries higher copyright risk than text because image generators have historically been trained on large scraped datasets that included copyrighted photographs and artwork — and several major lawsuits center precisely on this issue. Before using AI-generated images commercially, check the license of the tool you are using, understand what training data it used, and verify whether it offers any form of indemnification. The differences between tools are significant and have real legal consequences for commercial work. Learn more about AI image generation tools and their safe use.

• Check the licence of your AI image generator before commercial use — terms vary significantly by tool and plan tier
• Understand what training data the tool used: scraped web data carries more legal risk than licensed data
• For high-stakes commercial work, prefer tools with IP indemnification (see Section 6 for the full comparison)
• Register human-created elements of AI-assisted works where possible — the composition, selection, and arrangement can be copyrightable even if individual AI-generated elements are not
• Never use AI to generate images that closely replicate a known artist’s style or identifiable work
• For images containing human faces, check the tool’s policy on likeness and model releases

For Business Use

Organizations deploying AI content at scale face additional requirements beyond individual creators. Your internal AI content policy should address copyright risk explicitly — including which tools are approved, what human review is required before publication, and how you document the creative process for IP protection purposes. Understanding how AI tools handle your data and content inputs is equally important, since some consumer-tier AI tools use your inputs for further training by default.

• Include AI content policy in your IP agreements, employee contracts, and client deliverable terms
• Disclose AI use where platform terms require it — Google’s search quality guidelines, Amazon KDP, and LinkedIn all address AI-generated content
• Keep records of human creative contributions to AI-assisted projects — these records may be essential in future registration or litigation
• Use enterprise-tier AI tools where possible — they typically offer stronger data protection and, in some cases, indemnification (see Section 6)
• Brief your team on the difference between AI-assisted content (protectable) and purely AI-generated content (not protectable)

🏢 5. What Businesses Using AI Content Must Know in 2026

For businesses — not just individual creators — AI copyright compliance in 2026 involves navigating guidance from regulators, platform policies, and the terms of the AI tools themselves. Getting this wrong carries real commercial risk: reputational damage, platform delistings, client contract disputes, and potential liability for copyright infringement. The good news is that clear rules exist in most areas, and the businesses best positioned for 2026 are those that have already built the governance structures to manage AI content responsibly.

The U.S. Federal Trade Commission has issued guidance requiring disclosure of AI-generated content in certain advertising and marketing contexts, particularly where AI is used to generate testimonials or product claims. The California AI Transparency Act, effective January 2026, requires platforms and publishers to label AI-generated content. This is separate from the EU AI Act requirement, which mandates labelling of deepfakes and synthetic media for EU audiences. Businesses operating across both markets must comply with both frameworks simultaneously. Additionally, businesses in banking and financial services must note that U.S. Federal Reserve guidance SR 26-2, effective April 2026, extends model risk management requirements to AI and ML systems — which includes AI content generation used in customer-facing financial communications.

Platform Rules for AI Content in 2026: Google’s Search Quality Guidelines flag AI-generated content that lacks original insight as low-quality. Amazon KDP requires authors to disclose AI-generated content in their publications. LinkedIn prohibits misleading AI-generated profiles. Violating these policies risks deindexing, account suspension, or delisting — independent of copyright law.

One of the most important business decisions in AI content management is which tools to use — and specifically, whether those tools offer any protection if their outputs trigger a copyright claim. This is the indemnification question, and the answer varies dramatically by tool and by plan tier. Understanding this distinction is not just a legal nicety; it should be a core part of your AI vendor due diligence process before onboarding any AI content tool for commercial work.

Licensing terms as of June 2026. Verify current terms directly with each vendor before commercial use. This table is for informational purposes only and does not constitute legal advice.

🔐 6. AI Copyright and Data Privacy — What Happens to Your Inputs

Many creators and businesses focus on copyright in their AI outputs — but an equally important concern is copyright in their inputs. When you paste a client’s draft, a competitor’s document, or your own unpublished manuscript into a consumer-tier AI chatbot, you are potentially exposing that content to the tool’s training pipeline. Most consumer-tier plans for major AI tools use your inputs to improve their models by default, meaning confidential or proprietary content you share with an AI tool can, in theory, influence that model’s future outputs for other users. This is not just a data privacy concern — it is a copyright concern too, because the unpublished content you own and share with an AI tool may end up in training data that another company holds.

The risk is most acute in three scenarios. First, when you are working with client materials that you do not own outright and whose confidentiality obligations you must observe. Second, when you are working with your own unpublished work — a manuscript, a screenplay, a proprietary research report — that you have not yet registered or published. Third, when you are working with code that contains proprietary business logic. In each of these scenarios, the advice is the same: use enterprise-tier plans with explicit opt-out from training, or use tools that contractually commit to not training on your inputs. OpenAI’s API, Microsoft Copilot for Enterprise, and Google Workspace with Gemini Advanced all provide these contractual commitments at the enterprise tier. Consumer plans generally do not. Understanding how shadow AI use in your organization creates data and copyright exposure is equally important for business leaders managing teams.

The intersection of AI copyright and misinformation is a growing concern for publishers and businesses. When AI generates content that reproduces or closely resembles copyrighted work, it can also produce plausible-sounding but inaccurate information that, when published, creates both copyright and reputational liability. AI-generated images can be used to create convincing fake content that misuses real people’s likenesses or reproduces copyrighted visual elements — and digital provenance tools like the C2PA standard are emerging as the infrastructure for verifying what is real and AI-generated. Organizations publishing AI-assisted content at scale should implement provenance tracking as part of their content governance workflow.

🤖 7. AI Copyright Decision Framework: What Should You Do in 2026?

The AI copyright landscape is genuinely complex — but the practical decisions most creators and businesses face are not. The framework below consolidates the key questions you need to ask before using, publishing, or commercializing AI-generated or AI-assisted content. It is designed to be used as a quick reference, not a substitute for qualified legal advice in your specific jurisdiction.

The 2026 consensus for most creators and businesses is a hybrid approach: use AI tools aggressively for productivity and drafting, but ensure human creative judgment is visibly present in the final work. Document your process. Choose tools with appropriate licensing and indemnification for commercial work. Stay current on platform disclosure requirements. And treat your AI content policy as a living document — the court rulings expected in late 2026 and 2027 will likely require updates. Our guide to building an AI content publishing workflow gives your team a step-by-step process to implement these principles consistently at scale.

One final point that many creators overlook: the copyright risk runs in both directions. You need to protect your own AI-assisted work — but you also need to ensure that the AI content you publish does not infringe on someone else’s copyrighted material. AI tools can and do reproduce elements of their training data in their outputs, particularly when prompted in ways that closely mirror specific copyrighted works. This is not a theoretical risk — it is central to the NYT v. OpenAI lawsuit. Checking AI output against known sources, using AI tools that have deployed output filtering, and applying editorial judgment before publication are not optional steps. They are the practical minimum for responsible commercial AI content use in 2026. For a broader view of where AI law is heading, our guide to AI regulation in 2026 covers seven major new laws reshaping how businesses use AI across all sectors.

🏁 8. Conclusion

AI copyright law in 2026 is in active transition. The rules that existed three years ago — loose, uncertain, and largely untested in courts — have been replaced by a growing body of rulings, regulatory guidance, and settlements that give creators and businesses real direction. The human authorship requirement is now constitutional bedrock in the United States. The EU’s opt-out framework is live but imperfect. The UK has stepped back from legislative reform and is watching global developments closely. And in courtrooms from New York to Munich, the question of whether AI companies can train on copyrighted works without payment or permission is being answered case by case — with no definitive resolution yet, but a clear direction toward licensing rather than free use.

For creators, the practical message is straightforward: use AI as a powerful assistant, add your own creative judgment and contribution, document your process, and choose your tools wisely. For businesses, the stakes are higher and the requirements more complex — governance policies, vendor due diligence, platform compliance, and sector-specific regulations all demand attention. The businesses that treat AI copyright compliance as a strategic priority today will be better positioned when the appellate rulings arrive. And they will arrive. The only question is whether your organization is ready when they do.

📌 Key Takeaways

🔗 Related Articles

• 📖 EU AI Act Explained: A Beginner-Friendly Compliance Guide + Practical Checklist
• 📖 AI and Data Privacy: How to Use AI Tools Safely Without Exposing Personal Information
• 📖 Digital Provenance Explained: How to Verify What’s Real Online
• 📖 AI Content Publishing Workflow: A Step-by-Step Process for Teams
• 📖 AI Regulation in 2026: 7 New Laws Reshaping How Businesses Use AI