(5 days, 16 hours ago)
Commons ChamberAgain, this is another false dichotomy being presented to us between opt in and opt out. That is why we have landed on the term “rights reservation”. A lot of the material out there is not copyright. That is either because it is long out of copyright—the law for most works lasts for 70 years after the death of the author or the first publication of the work—or because some artists have categorically decided not to retain their copyright. Tom Lehrer, the author of many satirical songs from the 1980s and 1990s, such as “The Vatican Rag” and “The Masochism Tango”, has deliberately surrendered his copyright.
This is a world where we want to make sure that the vast majority of rights holders, whether they be the record label, the individual photographer, the artist or whatever, have the right of control over their copyright—over whether it is used and how it is used—and if it is going to be used, they should be remunerated. I urge my hon. Friend, who I know has a great interest in this subject in his role on the Select Committee, to make sure that that false dichotomy between opt in and opt out is abandoned. We talk about rights reservation, because then, opt out might look remarkably like opt in.
In July of this year, it was revealed that 173,000 YouTube videos, including material created by globally recognised British musicians, news channels and artists, had been scraped into a dataset used to train AI models. Content from over 40,000 creatives has been found in this dataset, yet I do not believe that consent was sought from a single impacted creator to use their copyrighted works. It is clear that AI offers a fantastic opportunity for our economy, but it must supplement and grow industries rather than replace them wholesale. Creatives deserve to be compensated for their work. AI companies will happily pay the electricity bill for their data centres and wages for their staff, so why should they not also pay to access the creative content on which their models depend?
I completely agree with the hon. Lady. Of course those companies should pay for the content that they are using. I think she is referring to LAION-5B, which is the dataset that was produced in Germany. Interestingly, a court in Hamburg has decided that this is already covered by the exemption for data and text mining for non-commercial purposes for research. Subsequently, though, this has been used not just for research, but for other purposes, which is precisely the kind of area where there is a legal dispute. That is why we are trying to provide legal certainty in the UK as to what can and cannot be used, when it can be used, and how we can make sure that people’s creative rights are protected.