Rewriting the Script on AI scraping of intellectual property for musicians and other artists

2nd November 2026 - Recent PQs submitted - will upload responses when received

 To ask the Minister for Enterprise, Trade and Employment the steps he is taking to ensure that representatives from the Irish music industry, including IMRO, RAAP and the Screen Composers Guild of Ireland are legally mandated to hold board seats or advisory positions within the upcoming AI Office of Ireland; and if he will make a statement on the matter.

To ask the Minister for Enterprise, Trade and Employment, in light of evidence that over 120 songs by an Irish band were scraped by AI developers without notice or compensation, what compliance checks or audit frameworks the national competent authorities will enforce on generative AI firms headquartered in Ireland to ensure they compile and publish comprehensive summaries of their training datasets; and if he will make a statement on the matter.

To ask the Minister for Enterprise, Trade and Employment, following the recent European Presidency Associated Conference on 'Copyright and Human Creativity in the AI Era', what specific proposals Ireland is bringing to the Council Working Party on Copyright to overhaul the 2019 Digital Single Market Directive; if Ireland will advocate for replacing the highly burdensome 'TDM opt-out' mechanism with a mandatory 'opt-in' model requiring prior notification and licensing before any Irish creative work can be fed into an AI training set; and if he will make a statement on the matter.

To ask the Minister for Enterprise, Trade and Employment the diplomatic steps Ireland is taking at the Council of the European Union to establish a harmonised, legally binding collective licensing framework across the EU Single Market; how this framework will guarantee that Irish songwriters, composers, and performers receive fair, recurring royalties whenever their catalogs are utilised to train general-purpose or music-specific AI models; and if he will make a statement on the matter.

To ask the Minister for Enterprise, Trade and Employment to state the specific negotiating position Ireland coordinated on behalf of the EU at the World Intellectual Property Organisation Standing Committee meeting regarding AI and copyright; if the position included calls for global transparency standards regarding music data scraping; and if he will make a statement on the matter.

To ask the Minister for Culture, Communications and Sport if the Government will implement a policy ensuring that no public arts grants or state funding via bodies like Arts Council Ireland are allocated to projects utilising generative AI tools trained on copyrighted Irish music datasets where explicit, compensated rightsholder consent was not acquired; and if he will make a statement on the matter.

--

1st November 2026 - Introduction to why tougher legislation is needed

In July of this year, lead singer of The Script Danny O’Donoghue spoke out about the practice of AI data centres "sraping" 125 of the band's songs without permission and without paying any royalties. He again referennced this during a recent performance, highlighting a sinister practice which has seen over 21 million musical works being used in data training sets.

Following on from general AI "learning" tools, we now have AI music production models being sold at various subscription levels that purport to generate music from scratch and offer enthusiasts the opportunity to create musical works by inputting keywords. Except that they aren't generating music from scratch and cannot in any sense be said to be musical "works".

As someone who has released music in the past and hope to again, I am aware of the advances in technology over the years that allow songs to be created and recorded using digital audio workstations and pre-recorded VST plugins and help speed up the creative process organically. I'm also aware that there are AI programmes that pay musicians for their works and can be useful in generating ideas to embellish a human-created original song. However it is clear that the current norm is for musicians to be basically robbed by the tech companies -adding to the already poor royalties paid by streaming services - and that there remains inadequte regulation to protect those in the sector.

The same applies to writers and we have seen situations where books have been ordered in bulk by companies to scan into AI systems, again without paying the original creators of those works. This is not an educational issue involving inputting older works out of copyright, but rather the theft of human creativity. There are also huge concerns in relation to filmmakers and associated creative enterprises.

I am hugely concerned about the wholescale theft of artists' work. I have already contributed to a debate on this issue, speech below from the 14th of July, and will be tabling further parliamentary questions to the relevant Ministers in the coming days:

I welcome the opportunity to speak on the motion, not only as a Teachta Dála but also as someone who trained and worked as a journalist for many years creating original output and as someone who has had stints as a singer-songwriter, writing and recording original works, and who hopes to do so again in the future.

I do see genuine uses of AI as a tool to speed up your creative process and to generate ideas based on your own original human spark and ideas, but, to be clear, the dangers for artists vastly outweigh the benefits, and that needs to be redressed. What is happening brings the name of the band from the 1980s Pop Will Eat Itself to entirely new levels. AI bottom trawls for free, pays nobody their due reward and then regenerates original ideas into a mishmash that sounds plausible, reads plausible and looks plausible, were it not for the fact that the AI output is input-output. It is only as good as what it trawls through.

We have seen the rise of AI artists, AI music, AI authors and AI video generators. Admittedly, some of them are parodies but others are deliberate scams to outline the possibilities of the technology. Increasingly, things are being created for its own sake and the profits do not go to any original artists. Nobody can tell which aspect of a work is being used in the generative process, but it is guaranteed that the graft of artists is nowhere in it. What emerges are not the original ideas of AI, the programmers or the big multinationals that own these companies.

We have had a constant fight over decades to ensure creative work is valued as labour and not treated as a hobby or a free resource. Music artists, for example, had to put up with reduced mechanical and performance royalties from streaming services. Increasingly as Deputy Tóibín referred to, people have sought refuge in live performance. So far, it is the last retreat of human artistic endeavour, but for how long? It used to be that live music was to promote recorded works but now recorded works promote live music to a larger extent. If recorded works are being threatened by AI derivatives where will our future writers, singers, actors and film directors be? They will not exist. In light of that, I welcome the motion.

We in Ireland have always understood that art in itself has value. It is always a work or a craft; even if the process of inspiration can strike like lightening sometimes, it is often honed and edited over a much longer period. We cannot automate this lived human experience or our imagination. I read somewhere recently that our innate evolution as a species was due to becoming storytellers around the fire. People have spent years honing skills that can never be replicated by a machine, yet today we face this challenge that we would not have anticipated years ago.

Fundamentally, the economic foundation of our work as artists, including royalties, licensing commissions and performance fees, are being further eroded. This time, it is not the music industry or the publishing industry but AI and the big tech companies. The motion acknowledges the alarming findings of organisations such as the International Confederation of Societies of Authors and Composers in relation to the explosion of AI-generated streaming fraud and the failure of existing EU mechanisms, particularly the digital Single Market copyright directive of 2019, to provide meaningful control for the artists. AI developers can legally scrape and analyse copyrighted content for training unless the creator has opted out. I will discuss the difficulties to which that gives rise in a moment.

This was originally designed for search engines and data analysis tools, not generative AI, but AI companies now rely on it to justify the mass ingestion of original creative works. Transparency licensing, fair remuneration and accountability are the very least that artists are entitled to if those companies are going to trawl through their work in this way. It is slipping away, especially with the withdrawal of the AI liability directive. The latter would have given artists something we currently do not have, namely a proper mechanism to get compensation or redress when AI systems and the companies that own them essentially misuse our work or cause economic harm through reduced incomes. This directive would have introduced a rebuttable presumption of causality, meaning that if an AI system produced content clearly derived from an artist’s work, the artist would not have to prove every technical detail of how the system ingested, processed and reproduced that work. It basically states that courts may presume a causal link between the system's output and the harm caused, unless the AI developer can prove otherwise. It is a presumption of artistic integrity, and the AI company has to prove it did not steal the work.

The directive would also have required disclosure obligations, thereby allowing courts to compel AI developers to reveal things like training data logs and documentation, information that is currently hidden behind this corporate wall. In welcoming this legislation, I acknowledge that a lot of what we want to do is going to be difficult when it comes to enforceability. That is very common. The directive would have helped big time. The European Commission argues that the AI Act is enough in itself, but it does not offer any civil liability disclosure rights or anything else to win a case. The motion before the House basically calls that out and asks for measures to improve the situation, so I support it. We need infrastructure at EU level to help artists. Even if every word was taken on board and we started from where we are, we will have to follow a process in respect of what is already a race against time.

If we look at how the AI Act could be improved, for starters, there must be mandatory as opposed to voluntary transparency. That is in the code of practice. We need legally enforceable obligations requiring AI developers to disclose exactly what datasets they are using, what copyrighted works are included and how those works are processed. It is quite difficult, but it is technically possible. We need a European-wide registry of copyrighted works used in AI training. That would be the backbone of any licensing system. It does not need to be perfect on day one. It could start off with voluntary submissions, collective management organisations and the major publishers. It does need to start, and we should not listen to threats to the effect that companies will just pull their money and business out of Europe and move elsewhere. That does not wash. Europe is the sanctuary of basic rights. We should stand up for those rights. We also need any opt-outs to be technically enforceable. At the moment, artists can tell companies not to train AI on their work, but developers cannot reliably identify or remove that work from their datasets. The European Union should therefore require dataset tracking. We also need more powers in relation to collective licensing because individual artists do not have that power in the context of individual companies. IMRO, Recorded Artists Actors Performers, the Screen Composers Guild and their European counterparts can. The EU should establish a compulsory licensing framework such that if AI developers want to train on European creative content, then they should pay into a collective pot. That is how radio, television and streaming were made to pay dividends, even if we all argue that those dividends are too low.

As already stated, getting rid of the AI liability directive was a mistake. Artists need a clear path to compensation when their work is misused, where their likenesses are impersonated and where their income is undermined by fraudulent AI content. We could show leadership here. Ireland should use its Presidency of the European Council to push for a more targeted and effective directive. Performers should have explicit legal control over their voice, their face, their movement and their personal and artistic identity. I welcome the motion and hope that the Minister of State will take it on board.

ENDS