WHO OWNS THE OUTPUT OF GENERATIVE ARTIFICIAL INTELLIGENCE?
One of the most talked about use cases of artificial intelligence (AI) recently is generative AI – AI that can generate or assist with generating content. Generative AI can produce written content, images, audio and video. This content can be completely novel, or in a style of a particular person.
So, who generally owns content created by AI under UK law?
There is no clear consensus on this question, and it is likely to be discussed (and hopefully clarified) in courts and at government level as time goes on. As it stands, it generally depends on the extent to which the AI contributes and assists a human in the creation.
Under UK law, if the content is something that may usually be protected by copyright (e.g. original literary, dramatic, musical or artistic work, including software code), the AI system itself cannot be the owner of the works.
If the content is generated wholly by AI, under UK laws, the content may be copyrightable as “computer generated works” by the “person by whom the arrangements necessary for the creation of the work are undertaken”.
So the creators of the AI system, rather than the human user, or the business commissioning the content, is likely to be the owner of the content. In practice, difficulties may arise in determining who the creator/designer is in the content (it is likely different people are undertaking different “arrangements” throughout the AI supply chain) – this question has not yet been tested in the UK courts.
If the AI is used as an assistive tool in creating the content, it depends on the level of input of the human user and the AI system in the process. If the human user had minimal input, then the content is likely to be owned by the AI system creators, as above.
If the human user has more input and is only using the AI system as an assistive tool, the human user is likely to be the owner of the content.
Similarly, if you are creating something that may usually be patentable, AI systems cannot be ‘inventors’ and therefore content generated by AI cannot be patentable.
There may be other options for protecting the works generated by the AI under UK law, depending on what they are: if it includes a database, and it fulfils the criteria for database rights (there must have been a substantial investment in obtaining, verifying or presenting the data) then it may be protected in this way.
What to check in the contract?
DO NOT ASSUME THAT THE CONTRACT TERMS WILL BE IN YOUR FAVOUR!
Ownership
Who owns the generated content between you and the AI provider? Some AI providers take the position that the generated content is owned by the user (and therefore if the provider owns any IP in the content, it assigns it to you). Often, this is caveated by “to the extent permitted by applicable law” - to the point that the legal position is not clear. Some AI providers will instead licence the content to you. If so, you should consider whether this is sufficient for your use case, and what the restrictions of the licence are.
Third party IP infringements
What assurances does the AI provider give you in terms of third party IP infringement? AI providers should give you a warranty that the training data and use of the AI system doesn’t infringe third party IP, and offer an indemnity to back that up. It is likely that there will be many caveats/exclusions to these, which must be checked carefully, for example the generated content itself may not be covered, or use of the AI system combined with other products may be excluded.
THANKS FOR READING