Artificial intelligence (AI) and copyright

Generative AI and the use of ‘large language models’ poses many challenges to current copyright law and it is expected that new legislation and case law will clarify the rules in time. The key principles to bear in mind at the present time are that

1. only human beings, and neither non-human animals nor machines, can own copyright;
2. any copyright material can be used to ‘train’ an AI system only with the permission of the copyright holder.

An exception for ‘text and data mining’ for non-commercial research was introduced into copyright law in 2019, but this gives copyright holders the ability to opt out of these uses and the right to be afforded the opportunity to do so. Any system that has used copyright material without this permission/against the wishes of a creator who has opted out, is guilty of infringement and therefore anyone using that system to generate new work is a party to that infringement.

You cannot feed copyright protected material into a generative AI tool and produce ‘new’ content from it without permission of the copyright holder. The terms of use for AI tools often assign the responsibility for content entered into the program to the user. You must ensure that you have the permission to input and use the content in the AI tool. Please note that ICLA’s copying licences do not permit such use.