Please note, this is a script from episode five of our video series, the digital download: Generative AI, which you can access here.
In this episode, we're going to be looking at the IP issues relevant to the use of generative AI.
IP rights in AI-generated content
The question of ownership of AI generated work is not yet fully settled. Creators may be able to establish copyright ownership in AI generated content if it meets the requirements at law. For copyright to apply in New Zealand, there has to be a work, it has to be original, and there must be an author. For the work to be original, the user will need to demonstrate that they've applied sufficient time, skill and effort in creating it. Inputting a simple prompt is unlikely to be enough.
In many countries around the world, the author of a work is required to be a human. However, under New Zealand law, the author of a work can be a person that's arranged a computer-generated work. The more time, skill, and effort that the user inputs into creating the prompts and post-generation editing, the higher the likelihood that they'll be considered to have arranged the work and therefore be the author.
Organisations will also generally have all the rights they need to use output for internal business purposes, like producing strategy documents or preparing slides for a presentation. However, the position may be less clear for those wanting to exploit content commercially, or to stop others from using it. For example, if a business wants to use AI to help develop proprietary code, or wants to use images developed by AI on its website. This is because AI tools such as ChatGPT generate content based off the data they've been trained on, which may include works owned by others.
Managing the IP risks
Similarly, if you're jointly creating works using these sorts of tools, you could be considered to be joint owners of the IP and the outputs. In both cases, it's really important to agree a contract upfront which clearly sets out each party's rights to own, use, and commercialise the IP in the outputs.
For mission critical or high-profile works, the safer approach might be to use more traditional means in producing content until we get more clarity about the copyright ownership position.
That brings us to the end of our Generative AI Series. We hope you've enjoyed it and picked up a few tips along the way. Thanks for joining us.