Patents are often the go-to intellectual property rights for start-ups wishing to protect their innovations and start-ups in the artificial intelligence (AI) field are no different. Indeed, it is possible to protect innovations in this space with patents [i] but often there is a question of when is the correct time to file a patent application.
The development cycle of products which deploy AI can be long and expensive, often requiring several investment rounds and years of development whilst they hone their model. For this reason, start-ups in the AI space can often find themselves in the unenviable position of needing to file patent applications in order to protect their innovations ahead of an investment round. This may be the case before they have had time to finalise their modelling. This may mean that the start-up cannot produce data which shows that their invention provides the technical effect they say it does.
The lack of evidence to support inventive step may undermine the chances of successfully obtaining a patent as the respective patent offices may use the lack of data to deny the technical efficacy of the claimed invention. But how much of a problem is this? Can the plausibility of an invention be questioned on this basis?
Topping up the first filing
This problem can, to some extent, be remedied by filing what have become known as “top-up” applications in the first 12 months after filing and then basing overseas applications and both the earlier application and the top-up application. A top-up application may contain some more supporting data and information to support the invention disclosed in the initial application.
This strategy can be risky, however, as the later application, strictly speaking, does not benefit from the earlier filing date in respect of the added information. This means that any intervening public disclosures, i.e. those made public between the first filing and the top-up filing, may become relevant for some of the material in the top-up application.
Plausibility of inventions is a topic which has recently been discussed by the EPO (in G2/21) and it seems they may take some of the worry away from start-ups in the AI field who may not have completed the development of their model. G2/21 relates to insecticide compositions but also provides guidance applicable in other fields where they may be a long product development cycle, such as AI.
The EPO set out that:
“Evidence submitted by a patent applicant or proprietor to prove a technical effect relied upon for acknowledgement of inventive step of the claimed subject matter may not be disregarded solely on the ground that such evidence, on which the effect rests, had not been public before the filing date of the patent-in-suit and was filed after that date.” and
“A patent applicant or proprietor may rely upon a technical effect for inventive step if the skilled person, having the common general knowledge in mind, and based on the application as originally filed, would derive said effect as being encompassed by the technical teaching and embodied by the same originally disclosed invention.”
That is to say, when assessing inventive step of a patent application, the EPO may not reject the application because data was not available at the time the application was filed. This means that, in the event that data supporting technical effect is not available at filing, an application could still proceed without the data provided the skilled person could derive the effect as being encompassed by the technical teaching of the patent application having the common general knowledge in mind.
Helping out AI Start-ups
AI trained models have a very long development cycle and it may be commercially advantageous to file patent applications before the development has been completed. This is reality but not ideal as far as the patent system is concerned, where a sufficient disclosure of an invention is required.
The EPO are assisting start-ups in this instance by setting out that data which emerges afterwards may be used to support the patent application. That is to say, it is not fatal to a patent application if it is absent some of the data which supports the technical effect of the described invention.
Given that the question of sufficiency in AI patent applications is very open, we would definitely encourage innovators to provide as much information as they can in their early patent applications. This would apply to any technical field.
However, the EPO have suggested that an AI model need not be fully developed in order to form the basis of a patent application which will have a chance of being allowed.
Please contact us if you have been innovating using AI to discuss your innovation.