AI and Copyright

Photo: Media from Wix

Artificial Intelligence technology (AI), presents huge opportunities to transform our economy and people’s lives for the better, but it also presents significant risks. One major risk if that creatives across the world are having their work and livelihoods stolen by AI.

Here in North East Fife, we have a plethora of talented creatives, including author Marion Todd who recently contacted me about the fact that her work had been illegally uploaded to a site called ‘LibGen’. Meta then scraped the website for all its content to train its’ AI models. Not only is this a breach of copyright, it is also illegal. I soon realised that a chapter I had written for a book had also been subjected to this process.

During a debate in Westminster on the impact of AI on intellectual property, I raised Marion’s case and emphasised the need for transparency when it comes to where AI companies are sourcing the materials they are using to train their AI models. We need to see a full response from Meta to explain this and why the use of illegally uploaded material on LibGen was authorised in the first place.

I have also met with representatives from the News Media Association who represent many of the biggest news publishers in the UK and also convene the Creative Rights in AI Coalition to further discuss this.  

The Liberal Democrats have tabled five amendments to the Data (Use and Access) Bill which is entering Report Stage next week. The amendments would make sure that existing copyright laws are enforced on generative AI models, ensuring fair renumeration for works used to train these models. We are also seeking greater transparency from AI companies to ensure that creatives are empowered to know when and how their works will be used.

The creative industries have been clear that failure to apply existing copyright laws to AI model training presents an existential threat. They are being asked to allow their output to be used to train models that could be in direct competition with them.

The Government have proposed an ‘opt-out’ system with the expectation that if creatives do not opt-out, their work will be used to train AI models. Although it may sound reasonable, the EU implemented this several years ago and still have not found a model where this is properly functional. If the UK want to follow this, we need to ensure that it works fairly and gets the balance right.