Setting your IP strategy – Part I: Prioritisation of IP in AI

What to protect

 

The next step is to consider what will be most valuable to protect, at this stage, a number of factors should be considered.

 

1.Does obtaining IP protection for a product align with the wider business strategy?

If the protection cannot be used to further the business’ core objectives then it may not be a worthwhile investment.

 

2.When comparing the product to those which already exist, what is new about it?

Some IP rights are only available for products which have at least one new feature, for example patent and design rights.

 

3.How commercially valuable is the product?

The commercial value of a product is considerably more important than how impressive the product is. IP rights are unlikely to provide financial benefit if the new and improved system cannot be monetised.

 

4.If a competitor were to reproduce the product, would it be readily identifiable?

For example, if a patent protects an AI software which has a specific new feature, for a competitor’s software to constitute infringement of the patent it must be demonstrable that it uses the patented feature.

 

5.Are the details of the new product to be published?

Some IP rights will not be available if the material for which protection is sought is made public in any way or by anyone. Therefore, it is important to initiate the acquisition of any IP rights before publishing any details of the system.

 

Which IP rights to acquire

 

IP rights may be divided into two categories: registered rights (patents, registered designs, etc.) and unregistered rights (unregistered designs, trade secrets, etc.). In the AI space, the most useful IP rights are likely to be patents and trade secrets.

 

In order to obtain a patent, an application containing a description of the invention and how it may be reproduced must be provided to the relevant Intellectual Property Office (IPO), who will later publish the application. A patent also includes at least one claim describing the essential features of the invention, including at least one novel feature. The scope of protection afforded by a patent prevents competitors from making, selling, or using a product which exhibits each and every feature of a single claim.

 

However, securing patent protection may come with certain drawbacks. When applying for a patent, fees must be paid to the relevant IPO, along with additional payments for the services of a patent attorney, which can collectively amount to a considerable cost. Also, there is no guarantee that a patent application will be granted. In addition to this, a single system may include multiple inventive concepts; however, a single patent will only protect a single inventive concept. Protecting an entire system may require a number of patents, and therefore may increase the associated costs. Nonetheless, patent protection is regarded as a strong IP right in that for as long as the required maintenance fees are paid the owner of a patent can, for a maximum period of 20 years, prevent any competitor from making, using, or selling a product which falls within the scope of the granted claim(s).

 

When considering AI-related inventions specifically, patents may present additional hurdles. AI-related patents have been rejected or revoked in the past for failing to provide sufficient detail. Examples of such detail include the input data used in an AI model, the structure of an AI model, specific dimensions of input vectors, and the number of layers in a neural network.

 

Unlike patents, trade secrets  do not require registration, and so the associated costs may be lower. However, for information to be protected under the Trade Secrets Regulations 2018 (UK), (i) it must be a secret, (ii) it must have commercial value due to its secrecy, and (iii) considerable steps must have been taken to maintain its secrecy. If a competitor were to discover a trade secret illegally and either make it available to the public or use it for financial gain, the owner of the trade secret may sue them. However, a trade secret only protects against illegal acquisition of the information. A trade secret cannot prevent a competitor from producing a product protected by a trade secret when it has been independently conceived by the competitor, including reverse engineering.

 

When utilising trade secrets, it is important to keep track of what information is secret using a trade secret register. What is intended to be secret may change over time, and so a trade secret register should be kept up-to-date. Moreover, the relatively low cost does not mean trade secrets are to be used indiscriminately, as this could create confusion about which information truly requires confidentiality.

 

The benefits and the drawbacks of patents and trade secrets need to be considered on a case-by-case basis. Some of the key factors for consideration are the budget, the possible rejection of a patent application, the implications of public disclosure, and the strength of protection desired.

It is common that multiple IP rights can be used together to provide more effective protection. For example, a patent application may be filed covering the core tangible features of a novel product or process and the specific implementation details may be maintained as a trade secret.

 

First Steps

 

A good first step for a business is to develop an IP strategy document, outlining the business’ goals and how IP is being used to further them. The IP strategy document should be made available throughout the business to ensure that employees may consider how they can help the business achieve its targets and everybody can be on the same page.

 

For additional detail on the topics discussed, please refer to our recent webinar entitled Setting your IP Strategy for 2025 - Strategic Patent and Trade Secret Portfolio Development presented by Colin Paterson and Kimberly Bayliss.