AI and copyright survey

The Government has published proposals for reform of copyright law in the light of ongoing legal uncertainty around the use of copyrighted works for the training of (generative) artificial intelligence (AI) models – see here for full details.

Background

The Government contends that current arrangements are not working for both the creative sectors and the AI industry. Rightsholders are concerned that their works are being used to train AI models without their permission and that they have been unable to secure remuneration through licensing agreements. AI firms claim that there is a lack of clarity over how they can legally access training data and that this uncertainty risks stunting AI innovation within the UK.

In response to these concerns, the Government proposes to introduce a "text and mining exception" that would permit copyrighted works to be used in the training of AI models within the UK, subject to a rights reservation by rightsholders (i.e. an 'opt out'). In addition, the Government is proposing other changes, including certain transparency requirements to make it compulsory for AI tools and users to make disclosures about AI, for example, the sources of the training data and to identify works created using AI tools. The Government is also consulting on a possible change to copyright law that would mean that computer-generated works would no longer benefit from copyright protection within the UK.

The changes proposed above would bring the UK copyright position more into alignment with the current EU position on copyright and AI.

How to respond to this survey

TIGA believes that its members may wish to prevent other AI tools from being trained on their game's copyrighted audio-visual works, but may also wish to use AI tools themselves in its businesses in some cases.

TIGA would be grateful if you could respond to the survey questions below so that we can use the survey's findings to represent the video games industry’s perspective on the important subject of copyright law to policy makers. You do not have to answer all questions and may leave questions blank that you do not wish to answer.

Please reply to the survey by 5pm on Friday 7th February 2025.

PLEASE NOTE THAT THE RESULTS OF THIS SURVEY WILL BE ANONYMISED. YOUR INDIVIDUAL ANSWERS WILL NOT BE MADE PUBLIC, but the aggregated answers may be used as part of TIGA's consultation response.

What is the name of your company?*

How many people does your company employ?*

1-1011-5051-249250 +

Thinking about your organisation’s workforce, what was the trend in 2024 in comparison to 2023*

Employment increasedEmployment decreasedEmployment remained the sameDon’t know

Please provide a brief description of your company's core business activities (e.g. own-IP independent developer, work-for-hire external development, service provider, etc)?*

Part 2: The proposed text- and data-mining copyright exception

Under the proposed "text and data mining exception" ("TDM"), copyright owners would be required to opt-out their works from training or license them for AI training (and potentially obtain payments for their use). If the copyright owner does not opt out then it would not be a copyright infringement to train an AI tool on that work within the UK.

Please note that, even if the UK determined that it was not an act of copyright infringement to train an AI model on input data within the UK under the exception, it could still be an act of copyright infringement to make available the output data in the UK or another jurisdiction. IP laws are territorial in nature and this is a complex area of law.

Do you suspect that your company has experienced any copyright infringements within the past five years?*

YesNoDon't Know

Have any instances of copyrighted infringement related to the use of your IP as training material by AI developers?*

YesNoUnsure

What has been the impact of such infringements (please specify, for example, if this impact has been financial, reputational or otherwise)?

Under existing UK copyright and related laws, how easy is it for your organisation to enforce its IP rights in a practical way, including: (a) preventing instances of infringement; and (b) obtaining fair remuneration from AI models and AI users seeking to license your IP?*

EasyMixedDifficultDon’t know

Please add any additional comments

Do you currently ever permit your IP to be used as training data by AI tools?*

NeverSometimesOften (core aspect of business model)

Please add any additional comments

If the Government were to implement its preferred approach of an 'opt-out' rights reservation for rightsholders with a broad TDM exception (Option '3' in the consultation), do you anticipate that your approach to licensing IP to AI tools and users would change?*

I would be more willing to license my IP to AI tools and usersStay the sameI would be less willing to license my IP to AI tools and usersDon't know/unsure

Please add any additional comments

Does your organisation currently use AI tools in your business?*

Please add any additional comments

On balance, do you consider it more beneficial to your business to be able to use AI tools or protect your IP from being used by other AI tools?*

More beneficial to be able to use AI toolsMore beneficial to be able to protect IP from use by other AI tools

IP rightsholders in other sectors, such as music, sometimes license their rights to collective management organisations (CMOs), who are given a mandate to license their members’ works on a blanket basis. Would your organisation be willing, in principle, to join a collective licensing scheme with other copyright owners in the games industry to manage access to your copyright works and pay you a royalty for their use?*

YesNoDon’t know

Please add any additional comments

Should AI firms and other organisations implementing text and data mining be required to disclose the use of specific works and datasets*

YesNoDon’t know

Please add any additional comments

Does your view expressed in response to (12) above change depending on the purpose of the AI model that your IP is used for, and/or the size of the AI developer business that is using your IP?*

How should AI model developers and other organisations deploying text and data mining disclose the use of external works and datasets? (please specify)*

The Government is considering how the proposed 'opt-out' would work from a technical and practical implementation perspective. Which approach do you think the Government should take, e.g. through the use of metadata and robots.txt? Please provide the reason(s) for your answer. If you consider that different types of implementation are necessary in respect of different types of IP asset (for example, art, audio, text, code, etc.), please provide any suggestions as to how this could work.*

If not already provided in the response to (15) above, please explain which methods your business already uses (if any) to block the use of your IP by automated/scraping tools, including (if applicable) any comments or concerns regarding the ongoing costs of implementing such tools and how this may change in the context of the consultation's proposals.*

Do you think that the TDM 'opt-out' should have a blanket effect (i.e. covering all use of your IP by all third parties), or would you prefer to have different levels of control over different works and/or in respect of different use cases (e.g. search indexing and language training vs generative AI)? If you consider that different levels of control are required, please could you provide suggestions as to how this might be implemented practicably.*

For your business, how important is it for the UK to be aligned with EU laws on copyright and AI?*

Overall, which of the following policy options do you support?

No change: the Government should do nothing and retain the existing copyright and related laws as they are.Opt-in: The Government should strengthen copyright requiring licensing in all cases, meaning that AI models could only be trained on copyright works in the UK if they have an express licence to do so. In other words, AI companies would need to secure opt-in consent from rights holders.Opt-out (Government proposal): The Government should allow AI developers to train on material to which they have lawful access, but only to the extent that right holders had not expressly reserved their rights. In other words, rights holders would need to expressly opt out of the new legal arrangements if they do not want their materials to be used for AI training.Broad TDM exception: The Government should introduce a broad data mining exception to copyright law, allowing data mining on copyright works – including for AI training – without right holders’ permission and without the possibility to opt out of this.Other

Part 3: Copyright in computer-generated works

The Government is also considering removing the copyright protection currently given to 'computer-generated works' in the UK. Computer-generated work under current legislation means "generated by computer in circumstances such that there is no human author of the work". The legislation is old and pre-dates AI tools. As such, there is some uncertainty as to what degree of AI tool use would mean that there is no longer a human author.

The copyright author and owner of a computer-generated work will be whoever "undertook the arrangements necessary for its creation" under current law, which could potentially be the prompt engineer, the employer, the AI tool creator or someone else, depending on the facts and circumstances.

Please note that, even if a computer-generated work might be subject to UK copyright, it may not be protectable in other jurisdictions, for example, in the USA or in Europe.

Do you believe that works created solely by AI tools (for example, using prompts) should be protected by copyright in the UK?*

UpDownRemain the sameDon’t know

Please add any additional comments

Do you believe that works created with the assistance of AI tools in combination with other human input should be protected by copyright in the UK?*

YesNoDon’t know

Please add any additional comments

Would removing the copyright protection afforded to computer-generated works have an impact on your business? For example, would you currently rely on protecting in-game assets created with the help of AI tools within the UK as part of your IP strategy – even if those works might not be protected in other jurisdictions (for example, the USA)?*

Are there any significant areas of concern or emerging issues regarding copyright and AI that you think the Government may not be aware of (and that it should consider as part of its consultation)? Please specify:*