AI and IP: A lively topic at IBA Conference 2025

The U.K. High Court has ruled that Stable Diffusion, a generative AI tool that creates high-quality images from text descriptions is not an “infringing article” merely because it was trained on copyrighted images, finding that model weights do not store or reproduce training works. In their article, Michael Fekete, Mat Brechtel, Matt Diskin, Barry Fong and Sam Ip write that this decision will be closely watched in Canada, including for its guidance on individualized proof and class definition challenges in AI training cases. https://ow.ly/Ti8E50XosMT

This is an important decision for anyone dealing with IP and AI. It hits copyright issues, trademark claims, jurisdictional limitations, statutory evolution, and the many evidentiary challenges. But at 205 pages, I won't blame you for getting an AI summary... (after you read our article!)

The recent High Court judgment on Getty Images v Stability AI marks the first major UK case that tests how intellectual property law applies to generative AI. As Courts are called on to adjudicate matters involving complex technical issues, technical expertise carries increasing importance. One must clearly lay out factual and technical parameters of a matter, identify true points of dispute, and ground judgments in high-quality, robust evidence. At Keystone, we combine cutting-edge technical expertise, rigorous analytics, and deep industry insight. Our experience and secure technical infrastructure allows us to execute statistically robust experiments on AI models and algorithmic systems. Read more about the judgement here: https://lnkd.in/edWmwRi3

Tech companies are calling for wide access to copyrighted content to allow them to build the most powerful and effective generative #AI systems.

The AI gold rush just hit a $1.5 billion speed bump. A judge said Anthropic (aka Claude) could train its chatbot on books it bought—but not on pirated copies. Now they're paying $3,000 per book to settle. Turns out "move fast and break things" has a price tag. What this means for every AI company. Read the news here: https://lnkd.in/g9UJRerF #Anthropic #AI #CopyrightInfringement #FairUse #DisputeSoft

What do AI models depend on? Data. But what happens when the data you use isn’t data you own or have the rights to use in that way? Commercial LLMs seem like an amazing capability, offering users ways to ask questions and get answers without ever looking at the original source material. In a sense, they are a modern search engine, but instead of directing a user to the document that best matches their prompt, a document is synthesized with no link to the original material. This fundamental issue disconnects the owner of the material from the ability to earn from it, especially as most AI firms have been willy nilly with their ethics. But copyright holders, creators, and publishers are taking notice and putting the commercial LLM providers on notice. In some cases, they are already winning court cases based on this. The key to AI today is to focus on using it effectively and ethically. The legal risks of not doing so become great. https://lnkd.in/eWJCzbqc

AI isn’t replacing creativity — it’s reshaping how we define it. The UK High Court’s recent Getty vs Stability AI ruling is a huge moment for anyone working in content, media, or design. The court found that AI models like Stable Diffusion don’t store or reproduce copyrighted works, they learn from data, much like we do. To me, this signals a shift that’s bigger than law, it’s about mindset. The question is no longer “Can we use AI?” but “How do we use it responsibly and creatively?” For content creators and brands, this opens the door to a new kind of storytelling: where AI handles the heavy lifting, and human taste, insight, and intent lead the way. The playing field isn’t disappearing, it’s expanding. Worth reading this piece to understand where the creative world is heading: https://lnkd.in/dYg3_zVu

The role of Generative AI in traditional knowledge systems, becomes very important. Many of the traditional knowledge systems are society driven and it had a practice of transforming the same to the generations. These traditional knowledge systems so linked with the society. How about, it becomes the knowledge of somebody else and use that for individual benefits. What will happen to the owner of this traditional knowledge, are they get acknowledged in the realm of AI. This is something very interesting, the recent developments in Australia in this context make very sense. How the intellectual property rights get managed in the AI context. Which is very promising. The given article is worth reading in that context. https://lnkd.in/gd9Wn8KG Thanks & Regards, Vinod. 

The UK’s first “Copyright vs. AI” decision (Getty Images (US) Inc & ors vs. Stability AI Limited [2025] EWHC 2863 (Ch)) marks a clear win for the artificial intelligence industry. Read more in the blog post by Clive Gringras, Elisabetta Righini, and David S.: "The UK’s First Copyright vs. AI Decision: Key Takeaways on a Win for the AI Industry."

AI is such a complicated subject - the ethics, the environment, the future implications, the employment ramifications, etc. But it seems pretty straightforward to me that no one’s work should be stolen, whether to entertain or to educate.