Introduction
Welcome to the first quarterly Copyright and Generative AI (GAI) blog post of 2026. If you’ve come across this page, you’re likely working in the copyright field in some capacity. As such, the current debate surrounding what, if any, changes should be made to copyright law to account for advancements in GAI has likely piqued your interest. With so much information and so many opinions swirling around the internet from all sides of the debate, it can be hard to keep track of where the conversation stands. We hope these blogs will help make sense of it all.
Every three months, you can expect an updated post that includes important cases to watch, recent international developments, a summary of the latest discussions (with an effort to remain impartial) and finally, a suggested long-form article to help you dive deeper into one aspect of the subject.
So, without further ado, let’s get into it!
Cases to Watch
Thaler v. Perlmutter, Case No. 25-449 (United States Court of Appeals for the District of Columbia Circuit, 14 October 2025)USA
In the US, computer scientist Stephen Thaler appears to have reached the end of the road in his quest to register the work entitled A Recent Entrance to Paradise. Created autonomously by Thaler’s own GAI system dubbed the “Creativity Machine,” Thaler first filed a registration application with the United States Copyright Office (USCO) back in 2018, naming the Creativity Machine as the sole author. When the USCO denied the registration, citing the requirement for a human author, Thaler appealed to the US District Court for the District of Columbia and then the DC Circuit after his appeal was denied. The latter unanimously affirmed the district court’s decision, leaving only the Supreme Court as the final authority on the matter. Thaler’s argument was essentially that, since the Copyright Act does not explicitly define authors as having to be human, the USCO was making a mistake in not allowing the work to be registered in their database. Ultimately, the Supreme Court denied Thaler’s petition for a writ of certiorari, so he has essentially run out of options for further appeal. It is important to note that this denial means the Supreme Court did not, in fact, take the case and therefore did not rule on it, leaving the question of the eligibility of AI-generated works technically still open. It will be interesting to see if Thaler finds another route to push the issue towards a decision.
Gracenote Media Services, LLC v. OpenAI Foundation, 1:26-cv-01947 (District Court, S.D. New York, 10 March 2026)USA
In a unique case (thus far), a metadata company known as Gracenote has now entered the fray by suing OpenAI for copyright infringement. Owned by Nielsen, Gracenote is alleging that OpenAI used its “highly valuable” data to create its own “highly valuable” products, including ChatGPT, without permission. Importantly, Gracenote has emphasized the fact that by doing so, OpenAI is directly competing with and threatening its core business model by using Gracenote’s data to create products (or allow the creation of products) that essentially provide the same benefits and services as its own. As we have seen in the past, the market dilution argument is one that judges are particularly focused on when discussing the possible application of fair use, so this will definitely be a case to watch.
Like Company v Google Ireland Limited, Case C-250/25 (Referring Court: Budapest Környéki Törvényszék) CJEU, 3 April 2025 EU
Over in Europe, the Court of Justice of the European Union (CJEU) is currently hearing this case, which seems to be predicated on the assumption that the input of data in GAI models (i.e., the training) and the output of data (i.e., the content produced using prompts) are two separate processes that need to be treated differently in the case of international law. Google is essentially arguing that, since its models are trained in the US, it cannot be sued under other copyright law regimes (in this case, Hungary) for infringement in the training process because the work itself does not occur in Europe and therefore is not covered by European Law. However, when Advocate General Szpunar was asking Google to clarify the applicable law, he made a statement that seemingly challenges this assumption. He suggested that, in order to fully analyze any alleged infringement to determine where damage occurred, the court may have to analyze the entirety of the actions that led to the case at hand and the alleged infringement. In essence, he was proposing that the dichotomy between the input and output phases of the GAI process was unhelpful in copyright infringement cases. This interpretation could completely change the way these cases are handled globally moving forward, so we will certainly be tracking how this case resolves.
RTI S.p.A. and Medusa Film S.p.A. v Perplexity AI Inc., (Court of Rome, December 2025)ITALY
Also in Europe last month, the first Italian lawsuit regarding AI training was brought forward in December when RTI SpA and Medusa Film SpA filed a civil lawsuit in Rome alleging that Perplexity AI used their copyright-protected content without permission in the training of its large language model (LLM), Sonar. Similar to cases in the US, the plaintiffs allege that this unauthorized use of their works is not simply technical or internal, but rather constitutes economic exploitation of their works, which falls squarely under the rights of a copyright holder, according to both Italian and European law. The plaintiffs have clarified that any text and data mining exception would not apply in this circumstance; they have also referred to the Berne Convention's three-step test, which allows for exceptions to copyright law only in certain circumstances, provided the normal exploitation of the work is not compromised. Stay tuned for more updates on this landmark case!
As always, I want to share this excellent map of the GAI cases currently active in the United States. It’s the best resource I have found to date that summarizes the various (i.e., over 80!) cases currently before the courts in that country. Since so much of the world still looks to the US for guidance on developing new policies, you can bet copyright offices and lawmakers globally are keeping tabs on how all these cases play out.
These are just a few of the cases either currently before or recently decided by the courts in the US and elsewhere. It will be interesting to see how many of them feature in our next quarterly roundup in the fall!
International Developments
Something we're keeping tabs on here at Copyrightlaws.com is the various governments’ slow march towards a federal AI policy in various countries. In February, the Canadian government released the results of their October 2025 30-day sprint AI consultation in a "what we heard" report. Interestingly, the various contributions to the consultation were actually analyzed using AI, which has some analysts doubting the reliability of the analysis overall. No policy has been put forward yet as a result of this effort, but we are expecting to see movement on this in the coming months as well. The full report can be found here.
In February, US Senators Adam Schiff (D-CA) and John Curtis (R-UT) introduced a bill into Congress, entitled Copyright Labeling and Ethical AI Reporting (CLEAR). If enacted as is, it would essentially require that GAI companies submit a detailed report to the USCO of every copyright-protected work included in the training dataset for their GAI platforms. If the training dataset is publicly available, they must also include the URL for the dataset. The Registrar of Copyright would then have to establish and maintain a publicly available online database that would include each submitted report. The potential civil penalties for not adhering to this requirement would max out at $2.5 million without limitation to other copyright remedies. In the ongoing conversation about GAI transparency, especially in the case of training data, this Act could fundamentally change the current tendency towards black box models and enable a far better understanding of whose works are being ingested by these popular models.
While the Senators mentioned above are proposing an Act that would implicate all GAI models currently in existence, Senators Marsha Blackburn (R-Tenn) and Peter Welch (D-Vt.) are instead using their position to call on ByteDance CEO Liang Rubo to shut down the Seedance 2.0 product (a video production GAI model) until meaningful safeguards have been put in place to prevent further infringing outputs. Seedance 2.0 made an immediate splash online in March when it was used to create allegedly infringing outputs, including a new ending to the popular Stranger Things series as well as a fabricated rooftop fight between Tom Cruise and Brad Pitt. Unlike earlier video models, the content produced using this model is seen as so realistic as to compete with the Hollywood machine at a fraction of the cost, leading to significant backlash from Hollywood studios and unions. The company has since delayed its global rollout due to this pushback. Since this delay is unlikely to be permanent, we are curious how ByteDance will handle the possible copyright implications of its model’s outputs in the future.
The UK government’s report on last year’s consultation on copyright and AI was due on March 18th (more on that below), but ahead of its release, the House of Lords Communications and Digital Committee published its report on AI, Copyright and the Creative Industries on March 6th. This committee’s main recommendations included ruling out the text and data mining (TDM) exception with an opt-out model, figuring out how to protect artistic style and digital identity (the UK has no specific personality right or protection for digital likeness at this time), making transparency a statutory obligation, backing the development of technical standards, and providing for the creation of a fair and inclusive UK licensing market. The committee’s main point was that UK copyright law is sufficient as is, with the possible exception of a need for a personality right, and that GAI models should be assessed under existing law rather than updating the law to accommodate their needs.
Following the release of this committee’s report, the UK Department for Science, Innovation and Technology, and the Department for Culture, Media and Sport duly produced their long-awaited Report on Copyright and Artificial Intelligence by the mid-March deadline. While overall, the report seems to adopt a “wait and see” approach — that is, it does not recommend any changes to the law or policy at this time — there are some notable elements. Recommendations include that the UK government should not adopt the TDM exception with the opt-out mechanism approach it originally preferred. In fact, it dismissed all the original options suggested in favour of waiting for more data and evidence. However, the one area where the government appears keen to make changes relates to the possible introduction of a personality right (mentioned above), due to concerns surrounding digital replicas. Other than this statement, however, the government has opted to wait and see how the situation plays out in the courts, between businesses and in other countries. Considering the UK Society of Authors recently launched a logo to identify books written by humans rather than AI, it looks as though creators are unwilling to wait any longer for the government to make up its mind.
The European Parliament has also adopted a report to work towards recalibrating EU copyright for generative AI. Tabled by the Committee on Legal Affairs, this report focuses on the shortcomings of existing copyright law in the interest of making Europe a world leader in AI. Some recommendations include sectoral collective licensing for GAI training (excluding press and news media), enhanced transparency on the part of AI developers, extraterritorial application of copyright, efficient voluntary licensing, presumption of use by AI developers if transparency requirements are not fulfilled, and exploration of measures to address copyright infringements in outputs and the issue of deepfakes. While seen as extremely ambitious, critics are not sure it provides an effective roadmap for implementing these broad changes (including addressing potential technical difficulties). As the report has now been adopted, it remains to be seen how its recommendations will be put into action.
In Asia, South Korea's Presidential Council has put forward a 98-point AI action plan that seems to prioritize the needs of GAI companies over creators. This is surprising as, historically, South Korea has had a reputation for fiercely protecting its creative industries, so some critics have called this an about-face for the country's government. While not a full-fledged policy as of yet, this action plan suggests the direction the government plans to take as it attempts to remain competitive in the GAI space. Lots of updates internationally this month! And there are so many we did not cover here. As mentioned below, our monthly Copyrighting AI newsletter is the best way to get up-to-date information and links straight to your inbox. If this post was interesting to you but you’re craving even more, this is the newsletter for you!
Summary of Latest Discussions
The latest topic of interest in copyright circles regarding AI is whether AI prompts themselves can be considered intellectual property and therefore protectable under IP law, especially copyright law. While it is undeniable that prompt engineering has quickly become a desirable and marketable skill, the question remains as to whether detailed prompts can actually be protected by copyright. So far, the US Copyright Office seems to be pretty clear that content produced using GAI cannot be protected by copyright. However, could protecting detailed prompts be a way around this? Some argue yes, as they are human-generated and require creative effort. We'll certainly be keeping an eye on how this conversation develops over the coming months.
Another developing area of discussion is the prospect of voluntary licensing agreements between copyright owners and GAI companies. The most notable one, arguably, was Disney’s deal with OpenAI’s Sora model. This three-year agreement would have permitted OpenAI to bring more than 200 characters from the Disney, Marvel, Pixar and Star Wars universes to their app, allowing users to create prompted social videos using this designated IP while also enabling ChatGPT to produce still images. Notably, the licence excluded talent likenesses and voices in a nod to the current heightened awareness of publicity and personality rights. As part of the deal, Disney reserved the right to feature curated content on its Disney+ streaming service. In return, Disney agreed to support OpenAI with a large investment and integration of OpenAI’s products in Disney’s business operations. This agreement was seen as a potential path for GAI companies to take, at least with large rights holders, that could help them avoid costly litigation or possible precedential rulings against their argument of fair use. It will be interesting to see if licensing, rather than an update to copyright law, ends up providing a way forward. However, now that OpenAI has decided to shutter its Sora model, the licence agreement described above has been cancelled, leaving the feasibility of this type of agreement still very much up in the air.
Some important questions in this space are yet to be answered: How much human involvement is enough for a work to be protected by copyright and eligible for registration, where applicable? How can people document their contribution to works produced using GAI? What happens if a work produced using GAI is determined to be ineligible for copyright protection but is licensed anyway? And finally, how do we handle AI output which is obviously infringing? We may have some answers or suggestions of paths to take, but so much remains unknown. We’ll see how much changes between now and the next quarterly update!
Deep Dive
For this quarter’s deep dive, aside from the UK, EU and Canadian reports included above, we are suggesting Open Future’s Beyond AI & Copyright report for a look at how to develop a sustainable information ecosystem in the age of AI. Rather than looking at AI from the perspective of existential risk, this paper focuses on its impact on how we produce, access and consume information. It also discussed the need for public (rather than privately funded) AI models and how commercial actors might channel revenue back to all contributors to the information commons. Have a read— it offers some food for thought on the broader impacts of this GAI age.
More Information
That’s it for this quarter’s update. However, if you’re looking for more regular and up-to-date news about the nexus of copyright and AI, check out our free newsletter, Copyrighting AI. And if you have any thoughts on what we have shared here, please feel free to reach out!
