Skip to Main Content

Generative Artificial Intelligence and Copyright : Gen-AI Terms and Conditions of Use

A guide for University staff and students on the copyright implications of Using Generative AI tools.

How do the terms and conditions of use of Gen-AI tools impact how I should use them?

When using Gen-AI tools you should familiarise yourself with the terms and conditions of use, in particular what the terms say about who owns the output of using the Gen-AI tool, any restrictions on how you can use the output, whether the Gen-AI tool will reuse your input data (prompts) and what measures are in place for safeguarding privacy.   

Ownership of Output

Most Gen-AI tools like Midjourney and NotionAI assign all rights to the user and OpenAI states in their terms of use that “As between you and OpenAI, and to the extent permitted by applicable law, you (a) retain your ownership rights in Input and (b) own the Output. We hereby assign to you all our right, title and interest, if any, in and to Output.” It is important to check the terms and conditions of the Gen-AI tool that you use to determine where copyright protection lies. 

While ownership in the output may be assigned to you as a user you may be required to grant a licence to the provider of the Gen-AI tool to reuse the output.  

Reuse of input data (prompts)

It is important for a Gen-AI system to have access to data to learn, refine its outputs and improve the service it provides users. The extent to which data input (prompts) and outputs are reused varies between providers. You should be aware that when providing information as part of prompts you may be giving the provider permission to reuse your data. Some providers such as Microsoft 365’s Copilot, do not use user inputs and outputs to train the Gen-AI models or only do so if the user opts in (Anthropic Claude).  

Other providers expressly state that they may use inputs and outputs to train the Gen-AI model. For example, Perplexity’s terms of service states that the user retains ownership of both input and output data, but go on to state that “when you upload, submit, store, send or receive content to or through the Service, you give us (and those we work with) a worldwide license to use, host, store, reproduce, modify, create derivative works, ........ communicate, publish, publicly perform, publicly display and distribute such content.” While Midjourney assigns ownership of outputs to the user by using the services you grant Midjourney, a license to reproduce, prepare derivative works of, publicly display, publicly perform, sublicense, and distribute text and image prompts you input into the services, as well as any assets produced by you through the service.  

It’s important to be mindful of the data you share and understand the permissions you give when agreeing to terms and conditions. 

Privacy

It is advisable to be cautious when entering data into any Gen-AI tool as it is not always clear how that information will be used. It could be used to develop its underlying large language model (LLM) and potentially data could appear in the outputs of Gen-AI. These systems rely on large amounts of data to learn and make predictions which has raised concerns about the collection, processing and storing of personal data, how it is used and who might be able to access it. 

You can read the privacy policy of the service provider to see what data is collected. For example, by creating an account with one of these services, you will have automatically given them access to a considerable amount of data. The privacy policy of OpenAI state that they will collect your IP address, browser, settings, the type of content you view, the features you use, the actions you take, and the device and operating system used. They will also collect the contents of any messages that you send. They do have an opt out option if you do not wish to have your conversations used for training models. 

You should be cautious about entering sensitive and personal data such as health, sexual orientation, religion and financial information. The collection of this data in vast quantities can raise concerns about how the data is being used and who has access to it. Some providers clearly state that they use personal data for model training, but it is not always clear if they anonymise the data. Others are less clear about how users’ data is utilised. It is worth reading over the privacy policy of the service you use to remain aware of exactly how your data is being used and to understand that you can opt out or select certain parameters around what data can be collected.