How can AI raise privacy and copyright concerns?

In December 2024, the UK Government launched a consultation on AI and copyright.

The UK Government says creative industries and the AI sector are strengths, essential to drive economic growth. However, the two are not always happy fellow travellers. The widespread use of copyright material for training AI models has presented challenges for the UK’s copyright framework, and many think AI threatens creative livelihoods. Essentially, the tension comes from creative people wanting remuneration for their work, and AI data-hungry models needing input but not generally being inclined to pay for that input.

The last UK Government proposed a broad text and data mining (TDM) exception to allow scraping of copyright-protected work for any commercial purpose (including training of AI tools), without providing any option for rightsholders to opt their works out. This met with significant opposition. The matter has also been raised in the passage of the Data (Use and Access) Bill, where various amendments have been proposed in the House of Lords.

The new proposal, ‘TDM with rights reservation’, is an attempt to resolve and clarify the tensions in the current copyright regime. Rightsholders might argue that the law is not unclear and that the Government’s contemplated changes will simply help AI developers. Indeed, some say this is a licence for big tech to ignore copyright and creatives.

Key features in the proposal include data mining for any purpose, including commercial purposes, provided you have lawful access; it would not apply where the rightsholder has reserved their rights in relation to the work. Where they have done so, a licence could be required.

What does our current law tell us about the issues?

The current TDM exception under UK copyright law relates to non-commercial research. However, the EU is amending its TDM regime for the purposes of research by research organisations and cultural heritage institutions, and some now ask if the UK approach is too narrow.

The UK law is mainly in one statute – the Copyright, Designs and Patents Act 1988 (CDPA). Section 9(3) sets out as follows: “In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken …”.

Unlike many countries, Section 178 CDPA enables copyright protection in works generated by a computer in circumstances when there is no human author of the work.

The courts of England & Wales have a tradition of reacting to technological developments in a way that businesses can accept, and often faster than legislation can be passed, which has its advantages. This, as well as the UK copyright law’s recognition of ‘computer-generated’ works, stands in contrast to other jurisdictions. In the US, case law has established that an artwork autonomously created by a computer algorithm could not be considered an ‘original work of authorship’, because authorship is an entirely human endeavour and non-human expression is ineligible for copyright protection.

The English law case development dealing with unexpected questions of technology gives us an idea of how the courts might deal with things. An example comes from the 1983 decision in Brinkibon Ltd v Stahag Stahl where the buyers of steel were in London and the sellers in Austria. The courts held that the contract was formed in Austria because communication of acceptance was received by telex in Vienna. The case was much influenced by what businesspeople thought should happen, a viewpoint that legislation sometimes omits. Now we are seeing litigation on the use of AI reach courts faster than legislators consider how to address it.

In the context of AI and computer-generated content or copyright infringement by companies employing such technology, a number of things need to be considered legally:

• copyright, including copyright in images, text, and data mining;
• image rights (absent in the UK) and moral rights;
• privacy and personal data protection; and
• human rights and ethics.

A series of cases help us understand these issues: Irvine v Talksport Ltd (2003) gave some protection to image rights of the subject of an image. The High Court ruled in favour of Irvine, determining that he was entitled to control the commercial use of his image.

The Zeta-Jones v Hello!, Zeta-Jones v Daily Mail, and Rihanna v Topshop cases looked at breach of confidentiality, privacy, and copyright, interspersed with arguments under the Human Rights Act and the tension between the right to privacy and the right of free speech. The key point learned was that the Douglasses (Zeta-Jones) were entitled to protect their wedding’s confidentiality, so if an AI used images that can meet the confidentiality standard, there could be a breach of rights.

In Rihanna v Topshop, the court recognised the commercial value of a celebrity’s image, affirming that unauthorised use can lead to liability. So, if AI is used to generate an image of a well-known person from other images, in addition to potentially breaching copyright, the use of that image may amount to passing-off.

In the 2014 case Weller v Associated Newspapers, data protection and privacy interplayed, when photos in a public place of Paul Weller included his children; the judge ruled that you can still expect privacy in a public place, and that photos might constitute unlawful personal data processing. If AI is using images of young people, then this might be another angle of tension.

The law on personal privacy (per the European Court of Human Rights in the Von Hannover case) makes it clear that unless hypocrisy is an issue, people going about their daily business do have a reasonable expectation of privacy. It has subsequently been accepted that this extends to aspects relating to personal identity, such as a person’s name and a photo of them. Where an image is computer-generated but might be believed to be genuine, we must consider how to be fair to those whose rights are impacted. We can see that the courts have been willing to right perceived wrongs; and where AI is drawing on other creative works for its own creations, it is perhaps likely that the courts will find an infringement of copyright. Where AI uses and/or generates images of individuals, the courts may conclude that this is unlawful processing of personal data under the Data Protection Act, or a breach of privacy and respect for private life under the Human Rights Act.