The EU AI Act: content creators and copyright

Recent reporting of the EU AI Act (the Act) has reiterated the importance of the UK Government choosing the right path in relation to regulating AI companies and their Machine Learning Models, and in doing so affording fair protection to content creators.

The Act is the first legal framework on AI. It addresses the risks of AI and positions Europe to play a leading role globally. The Act assigns applications of AI to three risk categories:

1. applications and systems that create an unacceptable risk,
2. high-risk applications, and
3. applications not explicitly banned or listed as high-risk are largely left unregulated.

Media reports have said that the law was never conceived to deal with generative AI models: systems that can generate text, images or music with a simple text prompt. This will undoubtedly cause concern amongst content creators, as their copyright work can be freely accessed and copied with little protection in place.

Generative AI and data scraping

There is an ever-widening opportunity that was naively left open in the Act by lawmakers, who did not anticipate the advancement of Generative AI models, that use indiscriminate data-scraping technology to “learn” from online copyright material. The Act allows an exemption for “text and data mining” which was anticipated to assist educational resources, or similar instances of public benefit – but it was not supposed to be a legal loophole which would allow AI companies to infringe copyright and use content to develop competing content. The fact that this is an EU law comes as small comfort to UK content creators; the nature of the global digital marketplace means that their content can be accessed within any jurisdiction and freely used as training for Generative AI platforms, no matter where they reside.

It is now clear that AI technology has surpassed initial expectation, and there is an undeniable effect on content creators: AI models that are trained on literary, musical, or visual art content will be able to create directly competing content, to the detriment of human creators. And, this competing content will be created so quickly, and at such low cost, that competition is inevitable. The end result will be fewer jobs, less creativity, and a very much poorer society in every sense of the word.

What should the UK Government be doing?

Where the EU leads, the UK often follows. However, here in the UK we still have the opportunity to lead as a safe haven for content creators. Our current copyright law provides ample protection for content creators, if only the Government would double down and ensure its proper application and enforcement – together with insisting on a transparent reporting structure on the part of AI companies in terms of where and how they train their models. Currently proposed systems involving “Rights Reservations” (essentially, opt-outs) won’t work: they rely on content creators being proactive and having the requisite knowledge to implement their own systems, and in any event, it’s next to impossible to know every single location where your content resides online, particularly in relation to downstream copies.

Creatives in the UK will be waiting patiently to see how the Government will manage this issue. The current Government Consultation on AI closes at midnight next Tuesday, 25 February 2025, and it is not hyperbole to suggest that this might be the last opportunity to ensure the protection and celebration of our rich creative landscape, in the face of an unrelenting tide of copyright theft on the part of AI companies.

If you have questions or concerns about AI and the copyright of creative works, please contact Christopher Gabbitas.