When thinking about AI it is likely that patents will be the first form of protection which comes to mind, and rightly so, but it is important not to forget the value of other forms of IP to protect your AI business. A strong portfolio of a variety of different kinds of IP rights is likely to be necessary to provide you with robust protection against the various types of possible infringement you might encounter. In this article, we look at how trade marks can be used to protect AI and some of the difficulties in preparing a trade mark application for an AI business.

How can trade marks protect my AI business?

Whilst trade mark law is never going to be suitable to protect the functionality of your AI powered invention, it is likely that you will be using a trade mark to refer to your AI offering. This could be as simple as the use of your company name to offer AI development services to a third party, or if you have developed a sophisticated AI machine you might give that machine a personality of its own via the use of a trade mark, for example, IBM’s pioneering AI machine is referred to as WATSON. IBM owns numerous trade mark registrations for WATSON covering computer hardware and software, data access services and various computer services, amongst others. It is important to protect this name using trade mark law to prevent third parties imitating your business by offering similar services under a mark which is the same as, or similar to, your chosen mark.

What are the main challenges in registering trade marks in the field of AI?

The key challenge trade mark owners face is preparing a suitable specification of goods and services. The specification will define the scope of protection granted to any resulting trade mark registration and it is therefore important to get it right. There are a few key challenges in this area.

The first is that AI is a rapidly developing area and it is likely we will see significant advancements in this filed in the coming years, but how do we include in a trade mark specification a technology which might not yet exist, or might not yet have a commonly accepted name?

The only term appearing in the current version of the Nice Classification (the classification system used in many countries, including the UK and the EU) relating to AI is ‘humanoid robots with artificial intelligence’, but this term alone is a long way off covering the full scope of AI machines already in production, never mind those which might appear on the market in the future. Of course, the lack of an official term appearing in the alphabetical list of the Nice Classification does not prevent the inclusion of other terms within a trade mark specification and it is promising to see tools such as TM Class including a number of ‘artificial intelligence software’ terms which should be accepted by the EU IPO and many other EU countries.

The second challenge relates to how broadly the specification should be drafted. There is a balance to be made between drafting a specification which is sufficiently broad that it is, to a degree, future proof and will encompass possible future advancements, but not so broad that the term is vague and imprecise. This could be a particularly pertinent point in light of the anticipated decision in Skykick which could change the current EU practice on broad terms such as ‘computer software’, and the issue of overbroad terms already exists in many other countries including the USA and China. You also need to be careful to use every day terminology which third parties and Examiners will understand to describe your goods and services as the use of technical terminology is likely to lead to objections if it is not understood by the Examiner.

Finally, it is important to think carefully about the goods and services being offered under the mark. The development of AI is likely to see a shift in the core offering of many businesses and it is important this change in focus is covered by the trade mark application. If we think of Amazon as an extreme example, traditionally its business was as an online retailer, but the development of AI has seen a vast expansion of its business into a wide range of different industries including the development of Amazon Web Services and Amazon’s own personal home assistance technology. A more relatable example might be your fridge, or other home appliances, automatically ordering goods, such as milk and eggs, when you run out, thus fridge manufacturers might suddenly find they need to cover new personal assistant type goods and services. Relying on pre-existing lists of goods and services could lead to a gap in the protection offered by trade marks and owners could find their go-to specification doesn’t cover core aspects of a new offering.