It’s safe to say that most sectors will undergo significant disruption as a result of artificial intelligence (AI) technology. AI will not only disrupt our business models but it will also disrupt our legal framework for the creation and exploitation of intellectual property (IP) rights, giving rise to new IP challenges for those seeking to develop and deploy new AI systems.
Ownership of AI generated content
With AI systems already being used to generate content capable of attracting IP protection, working out exactly who owns the IP rights in this content will become increasingly important, especially when it comes to licensing or enforcing those rights.
Ownership of copyright
Ownership of copyright is determined with reference to the “author” of a work. When an AI system is used by a human solely as a tool for creating a work, the same human using the system will generally be considered “the author”. But what happens when the AI system is more involved in creating the work, e.g. it is given a few simple inputs and the system goes on to create something which is much more than the sum of the inputs?
In the UK for example, the legislation states that where a work “is generated by a computer in circumstances such that there is no human author of the work” the author will be “the person by whom the arrangements necessary for the creation of the work are undertaken”. This provision doesn’t leave room for the AI itself to be considered the author. Instead, it sets up a two-stage process to identify the author; first, look to see if there is a human author of the work and, if one can’t be found, look for the person “by whom the arrangements necessary for the creation of the work are undertaken”. However, it is likely that disputes will arise at both stages when it comes to works which are generated by an AI system. For example were the inputs sufficiently related to the outputs to classify the person who provided them as the author? And who made the arrangements, was it the person who built the system, the person who trained it, or the person who fed it these specific inputs?
In order to apply for a patent, the application needs to identify one or more named inventors, with the common understanding that the inventor(s) are human beings. This requirement is not currently a major hindrance as, although AI systems are increasingly used to assist in R&D, there will almost always be a human who was actually responsible for devising the inventive concept. This could be someone who developed the AI system, defined the problem for it to solve, or reviewed the output to identify anything potentially patentable. There could, however, be a dispute between these individuals as to which of them is entitled to the invention, especially where they are not employed by the same organization.
How to manage the ownership of AI generated content
The Development of AI systems
When developing AI systems, all the usual IP issues associated with developing software products come into play, e.g. defining who will own the IP in the product (and any subsequent improvements), and who else has a right to use these. But AI systems also come with additional IP challenges.
Ownership of AI systems
AI systems will contain large sections of code which has been generated automatically, as a result of the system’s training. This differs from a traditional software development situation, where every line of code is attributable to a human author. So those parties entering agreements to develop AI systems need to think about who will own the IP rights in the trained system and how they will be licensed. Existing models based on who wrote the relevant section of code may fall down in certain situations.
Training the AI System
Developing an AI system generally involves training it using large datasets, so the system can continuously improve its decision-making abilities. However, it’s important to consider who owns the IP in the datasets which are used to train the system. One common misconception is that data which is available for free online can be used for any purpose. But this generally isn’t the case; website terms and conditions will often state that data cannot be used for commercial purposes, and using data without permission could also be a breach of third party copyright or database rights. In fact, using data without permission could endanger the future commercialisation of the system, at least until an alternative non-infringing data source is found.