Artificial intelligence (AI), and more specifically generative AI, is enabling people to perform tasks much more easily.
AI tools, such as OpenAI’s ChatGPT and Microsoft’s Copilot, are trained to source, or as it is known “scrape”, vast amounts of content from the internet to enable their users to create material in a matter of seconds, from student essays, computer programs, poetry and song lyrics, to music compositions, images and film.
Some of us are concerned. Teachers and employers are fearful that skills will be lost by pupils and employees. Sir Elton John, being one of a number of high-profile artists worried about “Big Tech”, is worried that AI developers will be “committing theft; thievery on a high scale” [source: BBC, Sunday with Laura Kuenssberg] if AI continues to use rights holder’s works without permission.
The UK Prime Minister Sir Keir Starmer is determined to press on regardless with the introduction of a series of policies on artificial intelligence, saying recently at London Tech Week that, whilst acknowledging the “social fears”, “By the end of this parliament we should be able to look every parent in the eye in every region in Britain and say ‘look what technology can deliver for you’ ”.
In order to press on, the UK Government has pledged a “Tech First” initiative, which includes a plan to teach school pupils AI skills. This may placate our teachers in part, but it certainly cosies up to Big Tech and will not satisfy the likes of Sir Elton. In order to implement these grand plans, the Government is continuing the previous Government’s review of the law to allow for the development of AI models without the fear of infringing intellectual property rights.
Copyright is the principal intellectual property right we are talking about here. Copyright in the UK protects original literary, dramatic, musical and artistic works, plus broadcasts, films, sound recordings and the typographical arrangements of published editions (which is the layout/appearance of a work).
As mentioned above, many generative AI tools source material from the internet, often without permission and sometimes without crediting the source. But does this infringe copyright, and who is the infringer?
In summary, copyright is infringed in the UK if the work (or indeed broadcast, firm, sound recording or typographical arrangement as mentioned above) is copied in any material form without the consent or licence of the copyright owner.
Underlining for emphasis, infringement arises where a restricted act is committed in respect of the whole or a substantial part of the copyright material, either directly or indirectly. The assessment of substantial is a qualitative, not quantitative, test, so copying a small part may infringe copyright, and copying does not necessarily need to be exactly the same or (in terms of words) verbatim.
Any person who has used generative AI will know that third-party materials are reproduced or copied in various ways by AI-generated content. But are these reproductions or copies restricted acts?
Copyright is not infringed by certain permitted acts (again underlined for emphasis), including copying for non-commercial research and “fair dealing”, which includes copying for criticism or review and reporting on current events.
AI could seemingly therefore be used to source content for non-commercial research, and both the user generating the AI content and the AI provider should feel comfortable that this is permitted use.
However, policing whether use is permitted or restricted is a tall order.
A lot of AI (if not the vast majority of it) does not discriminate and is trained to source certain content, whether copyright material or otherwise. Is it right to curtail the development of AI by putting the burden on developers to gain the necessary permissions from rights holders, but can the users of the AI be trusted to use the generated AI content for permitted acts only? The answer is, realistically, no. This is what the Government is grappling with.
The Government has been reviewing the possible extension of these permitted acts to encourage the development of AI and this is where Tech First conflicts with the artists and other rights holders. Certain pushbacks have already been made in respect of changing the law to permit text and data mining (TDM) by AI for commercial purposes, but consultations are ongoing.
Sir Elton is also worried that allowing the use of copyright works without permission will prejudice younger artists, who do not have the resources to challenge Big Tech. This could itself stifle creativity.
The issues are not only problematic for rights holders, though. Until the law changes, both the AI providers and the users of the AI face liability issues from those who can afford to protect their rights, such as Sir Elton, who has vowed to “fight it all the way”. The law remains untested, but a significant case of Getty v Stability AI is now in progress, which may provide guidance.
Under the current law, it will be very fact dependent on who will be responsible for any copyright infringement.
The general consensus at the moment is that the law hampers the providers of AI as they could well be liable directly if the AI models are trained to use copyright material without permission, or as joint tortfeasors when the user does a restricted act with the AI generated content, e.g. issues or communicates it to the public, or exploits it for any commercial purpose.
As alluded to in the last sentence, users of the AI themselves also risk liability under the current law. Although many people will be able to see that the AI generated content unfairly exploits copyright materials, many will be ignorant of the law which, of course, is no defence. The use of copyright materials in AI generated content may not always be obvious either, which fair attribution (crediting the ownership of materials) may solve if it is flagged to the user when the content is generated, and some AI does produce this as part of its generated content.
We will be monitoring forthcoming legal and policy development very carefully to see how the Government balances AI innovation with rights holders.
If you would like more information, please contact Kevin Sullivan at [email protected] or on 01732 770660.
This article is for general information only and does not constitute legal or professional advice. Please note that the law may have changed since this article was published. We do not accept responsibility or liability for any actions taken based on the information in this article.
