4 AI-Generated Content and Copyright Law

Navigating Ownership, Liability, and Legal Trends (2025 Update)

Artificial Intelligence (AI) is rewriting the rules of creativity, innovation, and intellectual property (IP) ownership. As businesses and creators worldwide integrate generative AI into their workflows, new legal questions arise: Who owns AI-generated content? What risks do AI developers and users face? How do US and EU copyright regimes differ?

This article provides fact-checked overview of the legal landscape in 2025, incorporating the latest court decisions, regulatory guidance, and practical compliance strategies.

Why Is Copyright Law Challenging for AI-Generated Content?

AI systems, especially large language models and generative algorithms, ingest vast amounts of data – often including copyrighted works – to create new outputs. While this expands creative possibilities, it also creates tension with IP protection.

Training data issues: Using copyrighted works for AI model training can lead to claims of direct or indirect infringement. The US Copyright Office’s May 2025 Part 3 report clarifies that some training uses may qualify as fair use while others will not, requiring case-by-case analysis.

Authorship debate: Legal systems require human authorship for copyright. The US Copyright Office’s January 2025 guidance reaffirms that purely autonomous AI outputs are not protected in the US, and similar principles apply in the EU.

Fair use and market impact: Courts must balance the transformative nature of AI-Generated content with the potential harm to original content creators, as demonstrated in recent high-profile litigation.

US vs EU – How Do Copyright Laws Compare for AI?

GEO tip: Targeting Europe and the US? Make sure your contracts and distribution policies align with local copyright enforcement and the EU’s emerging AI Act requirements.

Recent Court Cases Shaping AI Copyright Law

Thomson Reuters v. ROSS Intelligence (February 2025)

A US federal court ruled that using Westlaw’s copyrighted legal headnotes to train a competing legal AI system was not fair use, particularly given the commercial harm to Thomson Reuters. This precedent is significant for training data liability, though it applies to non-generative AI technology and is currently under appeal. The decision reinforces that commercial AI training on competitor content faces heightened scrutiny. ROSS filed notice of appeal in the Third Circuit in June 2025.

Anthropic Copyright Litigation Settlement (September 2025)

Following a split decision in June 2025, where a California court found that training on lawfully obtained copyrighted books could be transformative fair use but that using pirated copies constituted infringement, Anthropic reached a landmark $1.5 billion settlement in September 2025, which must still be approved by a federal judge.

This case illustrates the critical distinction between training on legally acquired content versus unauthorized copies, and demonstrates the financial risks facing AI companies.

US Copyright Office Guidance (January 2025)

The Copyright Office issued comprehensive guidance reaffirming that:

• Works produced entirely by AI without human creative input are ineligible for copyright protection
• Prompt engineering alone, regardless of complexity or sophistication, does not establish copyrightable authorship
• Human creative control must be exercised throughout the generation process for protection to apply

Copyright Office Part 3 Report (May 2025)

The pre-publication version of Part 3 on AI training and copyrighted materials takes a nuanced position: some AI training uses will qualify as fair use while others will not. Factors include the purpose of training, nature of the works used, amount copied, and market impact. This guidance moves away from categorical rules toward case-specific analysis.

How to Establish and Protect Ownership of AI-Generated Content

Design for human authorship: Ensure a human exercises meaningful creative control and makes substantive decisions at each stage of creation, not merely technical parameter adjustments.

Document contributions extensively: Keep detailed records of prompts, edits, selection criteria, and creative rationale. Simple prompts are insufficient – document iterative refinement and creative choices.

Edit and curate substantially: Combine, arrange, modify, and enhance AI outputs with significant human direction and original expression for legal protectability.

Register properly: In the US, make full disclosure of AI’s role and exclude purely AI-generated content from copyright claims. Misrepresentation can void registration.

Use comprehensive contracts: Clearly assign or license outputs and specify ownership, liability allocation, indemnification, and usage permissions in writing. Address both training data sources and output ownership.

Implement output review processes: Establish procedures to screen AI-generated content for potential copyright infringement before publication or commercial use.

Risk Mitigation for Businesses and Creators

Training data selection: Prefer licensed datasets, public domain materials, or content where you have clear usage rights. Document the provenance of all training data.

Vendor due diligence: Demand detailed information about training data sources, IP indemnities, and transparency from AI vendors. Ensure contracts allocate liability appropriately.

Output screening: Apply legal and technical checks to outputs before publication to identify potential infringement issues.

Insurance consideration: Evaluate whether existing professional liability or E&O policies cover AI-related IP risks, or obtain specialized coverage.

Monitor developments actively: Copyright law and AI regulation are evolving rapidly in 2025. Subscribe to updates from the US Copyright Office, follow key litigation, and track EU AI Act implementation.

Geographic compliance: If operating internationally, ensure compliance with both US fair use principles and EU opt-out/transparency requirements.

Frequently Asked Questions (Q&A)

Is AI-generated content copyrighted?

Only the portions of content where a human has made meaningful, creative decisions are protected by copyright. As confirmed by US Copyright Office guidance in January 2025, autonomous AI outputs and simple prompt engineering are not eligible for protection in the US or EU. You must demonstrate substantial human creative control throughout the process.

Can I use copyrighted works to train my AI legally?

It depends on multiple factors. The US Copyright Office’s May 2025 guidance indicates that some training uses may qualify as fair use while others will not. Key considerations include whether you obtained the works lawfully, the commercial nature of your use, and the potential market harm. The Thomson Reuters and Anthropic cases illustrate that using competitor content or pirated materials carries significant legal risk. In the EU, rightsholders can opt-out of text and data mining under the DSM Directive.

What happens if my AI tool outputs infringing content – Al-Generated Content?

Both the developer and the user can face liability. The Anthropic settlement demonstrates the magnitude of potential exposure – $1.5 billion in that case. Best practices include implementing output filtering systems, maintaining human review processes, obtaining robust indemnification clauses from AI providers, and documenting your efforts to prevent infringement. Consider whether your insurance covers AI-related IP risks.

How are 2025 court cases and regulations affecting the landscape?

Recent developments have significantly clarified the legal landscape:

• The Thomson Reuters case (February 2025) shows courts scrutinizing commercial training data use
• The Anthropic settlement (September 2025) demonstrates massive financial exposure for training on unauthorized content
• US Copyright Office guidance (January 2025) definitively rules out copyright for prompt engineering alone
• Part 3 guidance (May 2025) establishes case-by-case analysis framework for training data fair use

These developments emphasize the importance of human authorship, lawful training data sourcing, and robust compliance systems.

Do I need different strategies for the US versus EU?

Yes. While both jurisdictions require human authorship, the EU provides a text and data mining opt-out mechanism that the US lacks. The EU’s AI Act also imposes transparency and documentation requirements beyond US obligations. International businesses should implement policies that satisfy the stricter requirements of both jurisdictions, maintain comprehensive training data documentation, and ensure contracts address jurisdiction-specific compliance needs.

Conclusion

AI copyright law evolved dramatically in 2025, with landmark settlements, definitive regulatory guidance, and clarified legal standards. The Anthropic $1.5 billion settlement, Thomson Reuters precedent, and comprehensive US Copyright Office guidance all underscore critical compliance imperatives:

• Ensure meaningful human creative control throughout content generation
• Source training data lawfully and document its provenance
• Implement robust output review and filtering systems
• Use comprehensive contracts that allocate liability and provide indemnification
• Stay current with rapidly evolving case law and regulatory guidance

Whether your business operates in the US, EU, or globally, understanding your rights and responsibilities is essential to safely harnessing AI’s power while managing legal and financial risks. The cases and guidance from 2025 provide clearer boundaries than ever before-use them to build compliant, sustainable AI practices.

Sources used to cover Al-Generated Content and Copiright Law

Court Cases and Legal Decisions

1. Thomson Reuters v. ROSS Intelligence
   • Loeb & Loeb LLP: https://www.loeb.com/en/insights/publications/2025/02/thomson-reuters-v-ross-intelligence-inc
   • Davis Wright Tremaine: https://www.dwt.com/blogs/artificial-intelligence-law-advisor/2025/02/reuters-ross-court-ruling-ai-copyright-fair-use
   • Mayer Brown: https://www.mayerbrown.com/en/insights/publications/2025/02/ross-ai-decision-gives-early-indication-of-strengths-and-weaknesses-of-fair-use-defense
   • Thomson Reuters v. ROSS | BakerHostetler – https://www.bakerlaw.com/thomson-reuters-v-ross/
   • No One Can Own the Law’: Amici Come Out In Force to Support … – https://www.lawnext.com/2025/11/no-one-can-own-the-law-amici-come-out-in-force-to-support-ross-in-appeal-of-copyright-ruling-favoring-thomson-reuters.html
   • Headnote Wars in the AI Space: Thomson Reuters (Westlaw) v … – https://patentlyo.com/patent/2025/09/headnote-thomson-reuters.html
2. Anthropic Copyright Settlement
   • Official Settlement Website: https://www.anthropiccopyrightsettlement.com
   • NPR Coverage: https://www.npr.org/2025/09/05/nx-s1-5529404/anthropic-settlement-authors-copyright-ai
   • CNBC: https://www.cnbc.com/2025/09/05/anthropic-to-pay-1point5-billion-to-settle-authors-copyright-lawsuit-.html
   • Ropes & Gray Analysis: https://www.ropesgray.com/en/insights/alerts/2025/09/anthropics-landmark-copyright-settlement-implications-for-ai-developers-and-enterprise-users
   • Fortune: https://fortune.com/2025/09/05/anthropic-reaches-1-5-billion-settlement-with-authors-in-landmark-copyright-case/
   • Bartz v. Anthropic Settlement: What Authors Need to Know – https://authorsguild.org/advocacy/artificial-intelligence/what-authors-need-to-know-about-the-anthropic-settlement/
   • Anthropic Agrees to Pay $1.5 Billion to Settle Lawsuit With Book … – https://www.nytimes.com/2025/09/05/technology/anthropic-settlement-copyright-ai.html

US Copyright Office Guidance

3. Part 2: Copyrightability of AI Outputs
   • Official Report: https://www.copyright.gov/ai/
   • Crowell & Moring Analysis: https://www.crowell.com/en/insights/client-alerts/us-copyright-office-releases-part-2-of-artificial-intelligence-report-clarifying-copyrightability-of-generative-ai-outputs
   • IPWatchdog: https://ipwatchdog.com/2025/01/29/part-two-copyright-office-ai-report-says-creative-prompting-doesnt-constitute-authorship/
   • Skadden: https://www.skadden.com/insights/publications/2025/02/copyright-office-publishes-report
   • [PDF] Part 3: Generative AI Training pre-publication version – https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf
   • Artificial Intelligence Study | U.S. Copyright Office – https://www.copyright.gov/policy/artificial-intelligence/
4. Part 3: Generative AI Training (May 9, 2025 – Pre-publication)
   • Official Pre-publication Version: https://www.copyright.gov/ai/
   • Mondaq Analysis: https://www.mondaq.com/unitedstates/copyright/1627700/copyright-office-releases-pre-publication-report-on-copyrighted-works-in-generative-ai-training
   • Authors Guild: https://authorsguild.org/news/us-copyright-office-ai-report-part-3-what-authors-should-know/
   • Mayer Brown: https://www.mayerbrown.com/en/insights/publications/2025/05/united-states-copyright-office-weighs-in-on-fair-use-defense-for-generative-ai-training
   • Jones Day: https://www.jonesday.com/en/insights/2025/05/us-copyright-office-issues-guidance-on-generative-ai-training

EU Copyright and AI Law

5. EU Text and Data Mining (TDM) Exception
   • Guidelines and Code of Practice on transparent AI systems – https://digital-strategy.ec.europa.eu/en/faqs/guidelines-and-code-practice-transparent-ai-systems
   • High-level summary of the AI Act | EU Artificial Intelligence Act – https://artificialintelligenceact.eu/high-level-summary/
   • Code of Practice on transparency of AI-generated content – https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content
   • Transparency requirements re training data and compliance with … – https://www.hsfkramer.com/notes/ip/2025-09/ai-regulation-hasnt-taken-a-summer-break-transparency-requirements-re-training-data-and-compliance-with-copyright-law-come-into-force
   • CMS Law: https://cms-lawnow.com/en/ealerts/2025/01/interplay-of-ai-and-copyright-law-how-do-the-options-proposed-in-the-uk-s-recent-consultation-compare-with-the-eu-s-general-tdm-exception
   • Reed Smith: https://www.reedsmith.com/en/perspectives/ai-in-entertainment-and-media/2024/02/text-and-data-mining-in-eu
   • CMS (TDM Exceptions): https://cms-lawnow.com/en/ealerts/2024/10/ai-and-copyright-exploring-exceptions-for-text-and-data-mining
   • Morrison Foerster (Hamburg Court Decision): https://www.mofo.com/resources/insights/241004-to-scrape-or-not-to-scrape-first-court-decision
   • Kluwer Copyright Blog: https://copyrightblog.kluweriplaw.com/2025/04/14/eu-copyright-law-roundup-first-trimester-of-2025/