Introduction

Good evening all. My name is Anna Orchard and I'm a Chartered and Registered Trademark Attorney at Williams Powell here in London. I have experience in both intellectual property law and intellectual property law in the field of AI. I've completed publications in the Encyclopedia of Artificial Intelligence for both the first and second editions, which is due to be published soon.

My talk today will provide you with a high -level outline of what is IP, in essence, convey to you the need for intellectual property in your AI models and also provide you with a real -life case to understand the importance a bit further.

Intellectual Property Fundamentals

As some of you may know what intellectual property is, and it's simply a combination of legal rights that seeks to protect both your tangible and intangible products.

The main areas of intellectual property we'll be talking about today include copyright, trademarks, designs and patents. Some of these rights will be granted to you automatically, however some do require registration and applications.

Key Forms of Intellectual Property

Copyright

The first one we'll look at is copyright law. Copyright law is a form of intellectual property that is automatically granted to your work. Works in this case include your literacy works, your artistic works, musical, your software and databases, which might be important to some of you here, film and television recordings, just to name a few.

The main aim of copyright protection is simply to prevent someone from copying your work, renting it out or lending it or publishing it on the internet unknowingly.

As mentioned earlier this form of intellectual property does not require an application and is granted to you automatically as soon as the work is complete.

There is different lengths of protection and protection is dependent on the type of work. For example your written works is 70 years after the author's death whereas you have musical recordings for example are 70 years from when they're first produced.

you.

Patents

Well secondly we'll be moving on to patents. Patents are a form of intellectual property that protects your inventions.

A patent lasts up to 20 years as long as you renew it annually.

Please note an application for a patent is territorial. Your patent application in one jurisdiction or one country does not grant you protection in every single country so you need to be aware of that when making your applications.

You need to seek advice from local attorneys as well if you're looking to protect in different jurisdictions and looking to protect in more than one country.

If you're an inventor in the room and you wish to gain protection or you're entitled to the invention, you can apply for the patent application.

In order to meet the patentability criteria, your invention needs to be new, it needs to be novel, it it needs to contain some sort of inventive step and have some sort of industrial application.

There are certain things excluded from patent applications in the UK. For example, you cannot patent mathematical applications or a business plan.

It is a common misconception that you cannot patent software, and I know that's quite important for some of you in the room, but these can be patented in some jurisdictions and in some ways, for example, the US and the EU.

There have been hundreds upon thousands of software patent applications that have been processed but again that is something that you need to speak to a patent attorney to have the further knowledge on this.

Designs

The next area we'll look at is designs. Designs are a form of intellectual property that are most similar in a way to patents that we just discussed.

Patents protect as I just mentioned the invention whereas the design protects the actual look for example the colour the pattern of a product.

A design is is one of the shortest initial periods of protection as it only lasts for five years. You need to renew this and it can only go up to 25 years of protection.

In order for a design to be protected, very similar to patents, it needs to be new, but it also needs to have individual character. This means that it has to have a different overall impression to a different product that's already out there.

Designs can subsist and arrange features. For example, you've got your shape, you've got your configuration, you've got your pattern, your texture and your colour.

For those of you in the audience that are interested in applications, the user interface of an application can also be design protected, but you need analysis and checks from a design and pattern attorney in order to understand whether yours qualifies.

Trademarks

Last, but definitely not least, and arguably my favourite area of intellectual property, and I'm not being biased is trademarks. Trademarks are a form of intellectual property that protect your branding, your logos and your slogans.

Trademarks protect you for your services under what we call classes. They also protect your goods under these. These range from classes 1 to 45 and include terms such as software, artificial intelligence software for some of you, general formal clothing, all the way to business services.

After application, your trademark will remain enforceable for a 10 -year period after which you can renew it indefinitely.

As explained in the patent section, trademarks are territorial, so you need to provide an application in each jurisdiction that you're looking to protect in order to ensure you have that protection eventually globally, if that's what you so wish.

A trademark is quite advantageous as you can take action against those third parties that are highly similar to that of your own. It helps your company especially ensure exclusivity of your brand and that's extremely important when you're starting a brand or even in evolving and expanding your brand that's already established.

Getty Images v Stability AI

Now that we know the basics of intellectual property I'll dive into a case that's become very important in the field of IP and AI. Concluded in November 2025, the case I'll talk today is about Getty Images and Stability AI.

Now, I know a number of you in the room would have already heard of this case. This case in particular is very important to understand how intellectual property works with the ever -growing field of AI, especially on how these IP laws are put into play with generative AI models.

It brings to question whether the training and use of AI in the generative AI models constitutes copyright infringement, is a trademark infringement, have we got passing off? Are there database laws that have been breached? It also stresses the need to ensure that you have valid IP protection within your ever -growing AI world.

The Claims and Legal Questions

The case concerned Getty Images, alleging that Stability AI unlawfully used Getty's images to train its generative AI model.

The output results they alleged claimed that there was IP infringement on those outputs. The alleged infringement covered copyright and trademark laws that we'll be focusing on today. The watermark for example of Getty's images on the output was shown on the output results and therefore it was alleged claimed trademark infringement.

There are a number of claims that were made in the initial stages of the case that however were later dropped. These were dropped, for example, they had copyright infringement claims by Getty, but they didn't demonstrate the infringement in the jurisdiction of the UK. Now, this is extremely important that you have evidence of the infringement in the jurisdiction that you're looking to file the lawsuit. So in this case, it was a UK court, but they didn't have evidence or substantial enough evidence. It was claimed in the UK.

The copyright laws also brings into question the amount of human interaction is required for copyright law infringement to occur. It should be noted that copyright protection requires a level of human interaction and human authorship for copyright protections to be granted. Given the use of the AI models, the output results are technically, in essence, unprotected and presses the need for further assessments and governance in these areas. There's perhaps as well other areas of IP that we can consider in this case.

Getty also dropped the database claims as these were in some way linked to their primary copyright claims that I've just discussed. The case proceeded on secondary copyright claims and trademark infringement.

Secondary Copyright Infringement

Starting with secondary copyright infringement, this is in contrast to the primary copyright infringement, which is simply the act of copying. Secondary copyright infringement is more of a commercial infringement, where the defendant will have knowledge of the infringement occurring, whether they sell or deal or knowingly make copies of.

This area of infringement is extremely crucial and is extremely important to understand with the rise of AI as it is an ever -growing legal battle. This is because where AI models are trained, abroad and distributed, we need to really understand who has the liability and whether they even ever had knowledge of this infringement occurring.

In the event of this case, Getty needed to demonstrate that stable diffusion, the AI model in this case, was holding infringing copies within their artificial intelligence model. They needed to understand whether these were possessed in the course of business, whether they had been imported into the UK and or distributed in the course of business, which whether Stability knew or had reason to believe that this infringing copy was of Getty's.

The court however later dismissed these claims of copyright infringement, although an article under a copyright and designs patterns act can be intangible, a model like stable diffusion did not store any pixel data and therefore in short was shown that it didn't store the models by Getty images.

Stability was able to then prove that copyright infringement did not occur and that the model itself, the product of Getty's images was simply a product of the training and not itself a copy or an infringement.

Trademark Infringement and Watermarks

Moving on to trademark infringement, this case also related trademark infringement with the occurrence of an identical or similar sign and therefore Getty were claiming that the watermarks of Getty and iStock that they owned were produced in the output.

One of Getty's trademark claims was that identical or similar sign where a trademark has reputation or without due course takes unfair advantage of and is detrimental to the distinctive character and repute of the earlier mark.

The court dismissed Getty's claims here as they believe stability did did not intend to take unfair advantage of or take any infringement off Getty and they didn't show evidence that Stability's AI model used the sign and had caused them or had knowingly caused them damage.

Getty however had a very limited and narrow success in trademark law. The trademark victory concerned the display of the watermarks and the output of one of the iterations of Stabilities models.

So one of the older models of stable diffusion showed the output, showed the watermarks on Getty's images. These were labelled by the court as historic and extremely limited in scope, however it was a success.

Later models of stable diffusion did not produce these outputs, the watermarks, and therefore did not constitute infringement.

Why the Judgment Matters

This is the first UK judgment to analyse in depth how AI and IP and the crossover of these, and it's really important to understand understand the importance that IP has in this AI ever -growing world.

While some of the judgments here were either dropped or dismissed by the court it does still press the need you still need to look at IP law you need to look at trademark law for example whether you've got watermarks have these been protected and it's an important case therefore that I've highlighted today.

I know some of you in the room may be AI developers and this case is is important to you to understand the liability, who's to blame, who has the rights on, if there's any human interaction, who has the liability here.

I should press that proper legal advice is needed to ensure you understand the liability if you're creating an AI model, just to ensure that you have the correct protection, the correct advice moving forward.

Practical IP Considerations for AI

Whilst Getty Images is a landmark case for both AI and IP, we need to understand that there are other areas of AI that need to be understood.

Emerging Uses of AI

For example, I don't know whether you've seen, there's an ever -growing use of AI to recreate, resurrect actors from the dead. They're recreating them in films and the music industry and with this knowledge we need to ensure that we include IP in these areas and see whether or not IP can help protect the actors, whether it can be put into their contracts.

Conclusion

I believe the main point of this talk today is to ensure you're all aware of what IP is, why we think you should be considering IP protection within the early stages of your products, and especially when using or developing or embracing AI.

In short, IP protection can prevent third parties from replicating your brand, your brand name, your product. It can give you legal rights over your brand or your product.

It can strengthen your company assets, especially if you go on to later sell your company and ensure sure that you have that IP protection in place and it can heighten your brand recognition and exclusivity.

I hope my talk today has given you through to thought on the different areas of intellectual property and allow you to understand how intellectual property can be integrated into AI. Thank you for listening and I'm now available for any questions at the end.