Responsible and ethical use of AI

It is important to be aware of data privacy implications when using GenAI. Personal data must be protected in accordance with data protection legislation, and it is recommended that personal data is not used in GenAI tools. It is very important to check and understand the privacy and data policies of the AI tools/systems that you use.

In some cases the data you enter may be kept private and in others you may be contributing data for further training of the AI system or handing over information that the tool can store and use in other contexts. As a general rule do not put anything into AI that you would not be happy sharing publicly on the internet.

Since the training data for some AI systems draws on information broadly available on the internet and the AI systems typically use text, images, videos, sounds etc. indiscriminately in response to a prompt or a request there are ongoing debates about the intellectual property rights associated with content generated by AI systems.

The new technologies are in transition and transformation leading to gaps between current law and AI usage and practice. Staff and students should gain a basic understanding of the limitations imposed by copyright law so they can make informed decisions about using AI. In addition to observing copyright in your own use of AI you should ensure that students understand which materials they can legally upload to AI tools.

Key issues.

Who owns the copyright of material you are uploading?

Before uploading information into an AI system that will be training on or sharing the data you need to consider who owns the copyrights to the work you are uploading?

For example, uploading copyright material to AI could be breaching copyright or licence terms and conditions. University subscribed resources, e.g. e-journals, e-books typically can't be uploaded to AI tools. This is a legal risk that needs to be understood by all users if they are to use any Warwick-subscribed resources.

The only risk free material acceptable for AI uploading to AI that do not have strict privacy and data protections would be:

• where users own or have permission to use/upload the material to AI;
• material which is in the public domain;
• Creative Commons licensed material (excluding Non-Commercial use only and No Derivatives material);
• Out of Copyright material.

Who owns the copyright of the outputs?

In UK law if a work is generated entirely without human authorship it may still be protected as a “computer-generated work” in UK copyright law for 50 years from creation. For example, this could be in response to a simple prompt. Its author will be whoever undertook the arrangements necessary for its creation. This law was written for older computer programmes and so its application to AI remains uncertain.

There is debate about whether this balance within copyright law relating to the outputs of AI is right, between the traditional creator, the level of human input AI software might have, and the right holders in such works. These issues and Government proposals to address them are currently (February 2025) under consultation.

(Under current EU and US law material created by AI has no copyright protection since copyright requires an identifiable a human author, computer-generated work without human input also lack copyright protection in Australia, Canada and many other jurisdictions.)

Copyright infringement

If outputs created by AI are too close to existing copyright material then it could be considered a copyright infringement.

For example, if an AI system creates an image that substantially reproduces a specific Picasso painting (like closely replicating "Guernica"), this could infringe the copyright in that painting. This differs from merely creating art "in the style of" Picasso, which generally wouldn't infringe copyright. The key is whether the AI output copies protected elements from a specific copyrighted work.

Responsible use of AI includes consideration of how to maintain fairness and ensure that AI is inclusive and accessible to all. Everyone has different access to equipment, technologies, stable internet and AI tools. The university supports and provides some access to equipment and technologies, but when selecting AI tools for inclusion in teaching and study activities think about how to structure activities with everyone equally using free or paid platforms and equitable access to the tools required for the activity. For example, access to the internet may be crucial to complete an activity but the activity should be able to be completed using a slow or fast internet connection.

In the age of AI, we need to develop capacities to think for ourselves and to learn to think with AI and to learn to collaborate with emerging technologies. However, we need to also guard against over-reliance. For example, if one always uses an AI navigation tool to guide them from place A to place B then they may forget how to find their way around by themselves. Similarly, we need to engage with diversity in order to be better equipped to adapt when needed. We all have certain preferences but occasionally trying something different or expanding our comfort zones can contribute to our growth. Likewise we can take advantage of the conveniences and efficiencies of AI and emerging technologies, which we can calibrate to our preferences and ways of engaging with information but we should try to avoid becoming too narrow and too dependent on technologies. We have to actively mitigate letting AI think and work for us in order to maintain our human capacities and valuable contributions.