The consultation is now closed. Check our campaign page for further updates here
By Sandi Toksvig OBE
Who doesn’t want to hear about a success story? One of the UK’s biggest? It’s the creative industries which contribute a staggering £125 billion to our economy plus, of course, providing cultural discourse and enhancing the UK’s soft power abroad. What’s not to love? But all that risks being fundamentally undermined by the Government’s proposed changes to copyright law.
Copyright law! Not a top topic for everyone but it’s important – it’s what helps writers and other creatives control how, where and when their work is used and by whom. It fosters innovation and risk-taking and brings in money for writers through royalties and the Government through tax receipts and investment. It keeps us going so we can provide books, TV shows, plays, films and anything else written which might divert or entertain.
As things stand when one of these things is created the writers(s) who created it have the right to decide what happens with it. They can decide if it gets sold, licensed, shown to the public or shared in a variety of different ways. Writers need to expressly agree to these deals and can negotiate on their terms because… well, it’s their work.
But in its consultation on Copyright and Artificial Intelligence, the Government has stated that it prefers a very different model – one where writers, creators and other rights holders will need to reserve their rights, or ‘opt-out’, to stop AI companies from accessing the work and using it to train their models.
There are a whole host of issues with this. The first is the principle – that this is a fundamental change to how copyright works and it’s one that gives tech companies an incredible ability to take what they like, when they like, without informing or compensating the writers of the work they use to train their models. It gives AI companies, who are mainly based in the US, the ability to create systems based on the intellectual property of millions of creatives for their own commercial gain, without any requirement to share their profits. If you are with me this far you can see this is neither fair nor reasonable.
The next issue is a practical one – that there is no current system which allows for writers or other creators to effectively reserve their rights when it comes to AI training and it will surprise you to learn that the Government consultation doesn’t provide any clues as to how this may work. (Ed Newton-Rex has gone into great detail here to explain why opt-outs don’t work.) In short, it’s not always possible for creators to put in place the measures on every copy of their work to stop it from being scraped from online sources. Even if there are measures, the ones which are currently available are ineffective and can be circumvented by AI companies anyway. The EU has tried to implement this approach but been found wanting. The WGGB, as members of the Federation of Screenwriters in Europe and International Affiliation of Writers Guilds, have issued a statement of this here. Opting out simply doesn’t work.
There is very little trust that AI companies would abide by any opt-out requests, it is already well known that vast quantities of work have been taken without permission. The Atlantic have uncovered evidence of hundreds of thousands of books, including my own, being used to train AI. It has also found out that dialogue has been taken from thousands of films and TV episodes for the same purpose. There are legal challenges taking place, but these are likely to take some time to resolve.
If AI companies want to make use of the incredible output of UK writers and creators, they should have to seek proper permission and pay for it. Licensing deals can be done, and some already have, such as the Harper Collins deal. Allowing AI companies to take this data for free will likely kill off an emerging market which could otherwise provide a boost to the UK creative economy. Worse still, it will harm the creative industries through the loss of licensing deals and reduced investment. The Government want to pursue a growth agenda but have not provided a full economic impact assessment on what the consequences of these changes may be. Indeed, it would appear as though they are gambling the UK’s future on the potential of AI at the expense of an already successful creative sector.
What we really need is for the Government to protect and strengthen copyright and ensure that AI companies are made to play by the same rules as everyone else. We need the Government to introduce new laws which force AI companies to disclose what data they are training their models on and hold them to account when they have not received the necessary permission to use copyrighted work. The amendments to the Data Bill, introduced by Lady Kidron, and passed in the House of Lords, provide a blueprint for how this can be done.
As WGGB President, I would strongly encourage everyone to respond to the Government’s consultation, rejecting their preferred option and encouraging them to introduce proper transparency measures. The WGGB have put together a briefing which can help writers to fill it in, and the deadline is midnight on 25 February. Let’s use our actual intelligence on this and not some artificial stuff.