Home / Publications / Protecting AI brands

Protecting AI brands

Brand owners must be clear and precise when defining the scope of trade mark protection they require. A pending case in the European Court of Justice raises questions on this subject.
What are the implications for AI brand owners, and how can they be clear and precise about the trade mark protection they need in a field that is developing so rapidly?

Predicting the future

Artificial Intelligence Cloud

Companies developing AI technologies can make predictions about the product markets their technologies will occupy in five years’ time. However, how reliable and precise will their predictions be? For an AI company, technology development could easily lead them down new and unexpected paths, and product categories might expand from pure technology offerings into, for example, the automotive or medical fields.

As an AI company develops its technologies, attracts interest, and gains renown, its brand will grow in value. Any brand is a valuable asset. The ethical issues surrounding AI may mean that a trusted AI brand could have special value. As the brand would identify the core technology in which the trust reposes, its reputation would confer a significant advantage if extended to connected products for which the technology is used. Proper trade mark protection then becomes critical.

However, the inherent uncertainty about the future direction of its business can make it hard for an AI company to secure the most effective trade mark protection. The legal protection granted to trade marks is linked closely to the definition of the products and services for which a company claims the protection. It is easy enough to formulate that claim for a company whose business is established and relatively static (such as a law firm or a fashion label), but not for a business that is nascent and that may grow into or perhaps even create new product areas.

Future-proofing trade mark protection to deal with this kind of growth is particularly problematic for innovative AI companies. The concept of AI is itself the subject of rapid development and AI technologies are fuelling rapid convergence between technology and other sectors. Technology companies that
previously concentrated their trade mark protection on computer devices or software have subsequently expanded that protection to cover products such as driverless cars, healthcare and surveillance. These companies entered these new product markets by the developments in their AI technologies and the future that they see for them. Additionally, OEM’s interested in incorporating trusted AI technologies into their own products will be keen to secure a licence of the trade mark that goes with the technology and covers the relevant product product description. Their marketing will then benefit immediately from the trust attaching to the brand.

Forward-looking brand protection

Artificial Intelligence Arrow

The legal environment in which AI brand owners can secure robust trade mark protection is itself subject to a critical degree of uncertainty. In particular, a case currently pending before the European Court of Justice (CJEU) specifically raises the question of how far brand owners may legitimately go in securing valid trade mark protection of a scope wider than warranted by current commercial needs and concrete plans of the trade mark owner.

In comparison with the US trade mark system, the European system has traditionally been seen to offer greater scope for technology companies to secure adequate forward looking brand protection that is not pinned specifically to the products for which they have current use or verifiable plans. Unlike its US counterpart, the European system does not require an applicant for registered trade mark protection to specify the products and services it wants to cover with any particular degree of specificity. Nor need an applicant show actual use of its trademark for the products or services it specifies, or even establish that it intends to use the trade mark for those products. (In comparison, applicants for a UK trade mark have to declare an intention to use or licence the use of the mark applied for.) The EU trade mark system thus has the virtue that the owner of a new technology brand can opt for very wide protection at the outset.

Although the EU system does allow for the possibility that the extent of a trade mark’s protection would be cut back if it is not used for all of the products and services originally covered (the “use it or lose it” principle), that possibility arises only after five years. This gives the brand owner a substantial grace period within which to see how its technology and product offering develops. The brand owner may have to accept an eventual curtailment of its scope of protection. However, supporters of the EU approach would argue this is better than the brand owner having to try to cover new ground via new applications in future, having been constrained in the first instance to obtain a protection that with hindsight proves to have been too narrow.

The pending EU case concerns the enforcement of trade mark registrations which protect the trade mark in question for a very wide spectrum of disparate goods and services, covering around half of the 45 classes into which goods and services are classified. The coverage includes technology and telecommunications products and services of a broad description. In particular, it includes “computer software”, without any limitation by reference to the function or purpose of the software. The validity of these broad trade mark registrations are challenged on the basis that the trade mark owner could not conceivably have had a legitimate commercial interest in such wide trade mark protection and that the description of goods and services provided in the registrations lacks the clarity and precision which EU trade mark law requires. The point about lack of clarity and precision is levelled specifically at the unlimited reference to “computer software.” The question before the CJEU is whether or not these arguments amount to proper grounds on which the trademark registrations could be cancelled.

If the CJEU were to rule in favour of cancellation, this would significantly affect established EU trade mark practice, and could impact on a large number of existing trade mark owners. However, whatever the outcome of the case, it does raise some important issues: what does the EU law requirement for clarity and precision actually require in practical terms; what consequences (even if not the cancellation of the registration) would flow from failure to meet this requirement; and what does this mean for AI companies in particular?

Clear and precise

Artificial Intelligence Information

To understand this, it is best to start by considering why the requirement exists. EU law does not require clarity and precision in order to limit the scope of protection of trade marks to the narrowest legitimate compass. It requires clarity and precision simply to ensure that anyone consulting the trade mark register can understand what the trade mark registration does, and does not, cover. This is an important distinction, because it suggests that clarity/precision is not inconsistent with the use of descriptions that would encompass a wide spectrum of goods or services. On this logic, a trade mark registration should not be considered unclear simply because it describes goods and services in terms that are general in character, for instance “telecommunications services”.

On this basis, the use of the term “computer software” should not offend the requirement for clarity and precision even though there are many ways in which particular sub-categories of computer software could be identified, and which would have been sufficient to cover the products that the trade mark owner actually supplies.

A ruling by the CJEU supporting the use of “computer software” would come as a great relief to the tens of thousands of existing owners of EU trade marks that cover “computer software” without any kind of functional or other limitation. On the basis of the “use it or lose” it principle, the full scope of some of these registrations might well be cut back in due course, if it should be established that their owners have in practice used their marks only for a certain particular kind of software. However, this is separate issue.

The use of broad terminology covering a wide range of goods or services should not in itself be a clarity/precision issue. It could however be an issue if the terminology used is vague or fails to disclose what it aims to capture in terms of actual products or services (for example, references to “technology” rather than technology products of recognised kinds). In fact, concerns over clarity and precision are at least as likely to creep in when attempts are made to delineate goods and services in functional or other terms, within a broader category. This is because such delineation introduces product boundaries that, even if perfectly rational to the trade mark owner itself, may not be clear and purposeful to anyone else consulting the trade mark register. Also, limitations made expediently to trade mark specifications in an attempt to settle disputes with rival trade mark owners can themselves create uncertainty as to what is or is not covered by the resulting description. Sometimes the lack of clarity may not be immediately apparent at the time.


Artificial Intelligence Files

As for the consequences of this, the first point to make is that instances of lack of clarity will generally be picked up during the course of the trade mark application process. Trade mark examiners are required to consider this, and are adept at picking up uncertainties in the drafting. However, if the mark does make it through to registration, the consequences of a lack of clarity are that this will adversely impact on the enforceability of the trade mark. Where a trade mark owner relies, in a dispute, on a trade mark whose product/service description is not clear, whatever doubts exist on interpretation will be held against the trade mark owner. Thus, in cases of doubt, the narrower interpretation of the trade mark’s coverage will be chosen in preference to a broader interpretation. In some cases, the vagueness of the chosen wording may even lead to relevant parts of the trade mark’s coverage being ignored altogether.

Protecting an AI brand in the US and the EU

Artificial Intelligence Lawyers

Where does this leave AI trade mark owners? Technology brand owners based in the United States may look first for coverage on the US trade mark register. For that purpose they will need to follow the narrower approach to defining goods and services required under US practice. Having started out in the US, those companies may then quite naturally choose to adopt the same narrower approach when it comes to applying for trademark protection outside the US, including in the EU. However, even companies which start out looking for US protection do have other options when it comes to EU protection, and they may legitimately look for wider protection in Europe if they so choose.

Pending the CJEU’s ruling, a reasoned approach for AI brand owners looking to achieve reasonably broad and flexible trade mark protection in the EU would be:

  • Do not ignore the advantage of the European approach, allowing for wide registered trade mark protection to cover anticipated future uses of the chosen trade mark. This could include product categories that are not under immediate development, but into which use could extend within a few years. However, resist the temptation to extend the coverage into all kinds of product areas unrelated to logical developments of the brand.These are rarely crucial when it comes to meaningful brand protection.
  • Do not feel compelled to avoid broad terms like “computer software” simply because they are broad. If using this broader kind of product description, consider supplementing it with additional, more specific descriptions of particular types of software (or other product) that reflect the current core interests of the business or its immediate development plans.
  • Remember that being more specific does not necessarily make something more clear and precise. If looking to limit a product or service description by reference to the product’s purpose or function, or some other attribute, be sure that this does not introduce ambiguity or other uncertainty into the description.
  • Be prepared to make limitations to broad product/service specifications if the use of the trade mark over time remains limited in a way that does not justify the full coverage originally obtained for the mark, or if the CJEU should condemn the use of broad product descriptions.

Describing the core AI software

Artificial Intelligence Lawyers

The most central issue for AI companies would be the description of the core software. Under current practice at the EU trade mark office (the EUIPO), acceptable specifications include “artificial intelligence software” as well as certain more specific iterations, such as “artificial intelligence software for surveillance”. AI companies looking for product descriptions beyond the more general term “computer software” may therefore naturally look to this kind of description.

However, it remains to be seen whether such a description would actually survive a legal challenge since there is currently no widely accepted definition of AI. It is an open question whether “artificial intelligence software” allows for a sufficiently concrete distinction to be made between what is and is not covered. In the first place, notions of what constitutes artificial intelligence have developed significantly in recent years as the pace of technological change has intensified. There is debate as to whether AI software needs to be able to learn or not. There is also a lack of clarity because human intelligence is itself not well understood. Often AI practitioners use the term “artificial intelligence” to refer to a field of study rather than to a class of technology. Recently the European Commission has published a draft definition of the term “Artificial Intelligence” which includes a 7 page explanatory document and a disclaimer which apologises that the definition is “a very crude oversimplification”.

AI companies looking to specify their own software in some way should consider supplementing a description which uses the acceptable terminology with an additional description that avoids the use of “artificial intelligence” as a defining characteristic. The additional description should clearly and objectively describe the software, usually by reference to its nature and/or function. Language should be chosen that would be understood by competitors of the business, and that is not likely to be interpreted differently in five years’ time.

Artificial intelligence - Protecting AI brands
PDF 78.3 kB


Guy Heath