IPKat in conversation with Tim Moss, CEO of the UK IPO

This Kat was delighted to have the opportunity to chat with Tim Moss, Chief Executive Officer and Comptroller General of the UK’s Intellectual Property Office (IPO), on the topic of artificial intelligence (AI). 

Hayleigh Bosher: One of the IPO’s key priorities in its Corporate Plan is to understand the impact of future technology on the IP framework. What are the current capabilities of AI and machine learning and how do you see this developing in the future? 

Tim Moss, CEO UK IPO

Tim Moss: This is an incredibly fast-moving area of technology, and we are seeing ever increasing uses of AI, certainly we see it in terms of creating content, we are also seeing it in terms of AI being involved in the invention process. We’ve seen AI creating paintings, a very famous one Rembrandt, also in literature – I heard at the conference we did the year before – actual whole books being written by AI. Music is being created; Sony signed a deal for x number of albums created by AI. Often with a bit of human assistance, but AI is being hugely involved in this space. Recently there was a play called AI which was staged at the Young Vic, which used AI to create a different script each day, or co-created. So, it is being involved in the creative process; it is here, and it is going to ever expand. I think there are less examples of AI devised inventions, although there is the famous case of something called DABUS, which applied for patents with us and various other IP offices around the world. We didn’t grant it because of the big legal question around human involvement in invention, and it has gone to the courts to look at it. There are other examples of AI being used in invention, it is an incredibly important tool and that is going to increase further and further as things develop and it becomes more sophisticated. 

Hayleigh Bosher: We are going to talk about these big questions that we don’t actually know the answers to because they are undecided, but how do you see our current IP framework mapping onto the technology that you have just mentioned?

Tim Moss: The starting point is that we have a really great IP framework, and the best regulatory frameworks should be as technology independent as possible, principle based, but at the same time when you get things which ask some big fundamental questions, such as with AI and machine learning, you do then have to think about whether the framework is fit for purpose, now but also in the future. I think that is the really exciting bit that we are in at the moment, around the IP framework and AI. There are some really big and difficult questions, and we need to make sure we get this right. It’s that balance between making sure the framework adapts quickly enough to support these great and wonderful emerging technologies, but without being too sort of knee-jerk reaction; doing something without considering the long-term implications. Especially as all good frameworks should be as technologically independent as possible, because they are about the principles that sit behind it, but I think AI is certainly starting to question some of the principles and really changing the world that we are in. So, I think it is a really exciting time. 

Hayleigh Bosher: One of the big questions that you are alluding to, I’m sure, is about ownership. When AI creates something, is that something that someone can own? If the creation is something that qualifies as IP, who owns it? Could you speak on that issue a little bit and talk about what the IPO are doing in terms of looking into the framework and how it might be adapted. 

Tim Moss: You are absolutely right, the questions of authorship, ownership, or inventorship ais one of the biggest questions around AI and IP. The question is what is the link between human creativity and the output? Obviously, humans are developing the AI and humans are applying the AI, they are involved in it but that link between the human being and the output is starting to be stretched. And the question is, at point it stretches to the point when actually it no longer applies? That is one of the big questions. But, it’s not just about ownership and authorship, you are looking at the fundamentals of IP. One of the fundamentals of IP is that it provides that incentive to be creative. Does a machine or an algorithm need an incentive? But at the same time, on the other side of the argument, it is there also to protect people’s investment and to give them confidence that if they invest their time, their money, their energy, their technology, into creating something new, that they can then benefit from the rewards of their investment. So, you’ve got these really big fundamental questions around the IP framework, what it’s set up to do, its support for innovation and these huge questions are at the heart of what IP is about. So, as I said, that is where we need to tread carefully and get as much information from as many sources on as to how this is developing and that was part of why we did a call for views last year, we also did a big conference the year before, and we are looking to do another consultation this Autumn, looking at some more specific questions around how AI has impacted on IP. 

Hayleigh Bosher: One thing you mentioned, which I think is really interesting, which relates to some of my research in my book I looked at the different views of the music industry on what you are talking about, from a music perspective, and obviously there are different stakeholders; the incentive of the investor who create the AI and then there is a concern from the creators around whether AI-created content be valued and regulated in the same way as human-created content? Do you see a distinction there or a relationship between those two?

Tim Moss: Again, these are really key questions at the heart of it. Copyright has a rule, for example, linked around the lifetime of the creator, well an algorithm doesn’t have a lifetime. So, the current rules just don’t operate if you went down that path. So, having things that would apply and be suitable for a world where an algorithm completely independently creates something, if you had copyright linked to that then it just doesn’t work in the same way as it does currently for human-created content. Whether you can end up with different rules, or a situation where copyright doesn’t apply in certain situations and does in others, these are the things that we need to work out. And ultimately, it is taking that step back and asking; what is the purpose of the protection? And why is that important? And is it as important for something that is created by a machine as opposed to something created by a human being? And on the other side, you have got how do you enforce it? Copyright needs to be protected as well as granted, and who are you protecting for? As I said, we haven’t got the answers at the moment and that is one of the reasons for the consultation, and it would be great if readers want to comment on this and give their views because these are such important questions for the future of the system. 

The fundamental basis of the IP system is that it is a balance system. Essentially, you are creating a statutory monopoly for somebody and so there has to be a balance to that. Some of that balancing is happening in places around publication and transparency and also being time bound. But at the same time, you need to make sure you are balancing the needs of the public, business, inventors, creators so that the system works and works effectively rather than moving too far in one direction or another. We see our role as ensuring that the whole ecosystem works effectively, and that ecosystem is not just about just about IP, it is there to support the wider economy in terms of making the UK one of the most innovative and creative countries in the world.

Hayleigh Bosher: Well, I am glad you brought up the balance, as I have some questions around that. How do you envisage creators being remunerated for the use of their work in an AI algorithm, for example the input data? Do you have any thoughts on this, is this something that the IPO is looking into? And is there any consideration for a levy system being used in these circumstances?

Tim Moss: A starting point is that currently there are effective licensing arrangements in place and so if a work created by or used by AI, contains other copyright material then that material could be licensed to be used in that way. So, there is a mechanism that works currently. The question, where we are looking for evidence is, does that system get broken at some point, with the way that AI is using it? So, you can have an effective licensing regime, but if AI is trying to bypass it then does that fall into the areas of infringement, which goes into a separate question. Or actually is there some way of adapting the licensing system, whether it be a levying or things like that. So, at the moment we are not seeing anything that is saying that licensing couldn’t work, but that’s where we want evidence from people to show maybe areas where it doesn’t apply, or it couldn’t apply. So, if there are areas and rules where it doesn’t work, we might need something different such as a levying or a special licensing regime. But at the moment we think the system can currently work. 

Hayleigh Bosher: And on the other side of that, what about exceptions? Because the IPO and the Government have said about putting the UK at the forefront of innovation and I think that probably the AI developers would argue exceptions to IP at this point would help them to develop. But, like you were saying, it’s all about this balancing act. So, do you see any exceptions that already apply, or there needs to be new exceptions?

Tim Moss: This is certainly one of the big questions we will be asking as part of our consultation - whether we can make it easier for creative works to be used for machine learning and AI, it could be done through licensing, it could be done through exceptions – and we want to consult on the different options around that. I am certainly aware, from conversations I have had, about questions being raised about whether the text and data mining exception that currently exists is fit for purpose, given the importance of data to AI. I think that is probably one of the most fundamental ones; the role of data and training algorithms is where so much of the value is. We were one of the first countries to introduce a text and data mining exemption, now I know through the Digital Single Market Directive a lot of Europe is also implementing that and taking it slightly further. So, we need to be looking at that and saying is the current exemption fit for purpose or does it need to be adapted, especially in light of how data is being used? I am also aware of discussions, with WIPO and across the world, asking if there is a need for thinking about the rights associated with data in a different way, because of the way that data is going to be used in the future and it is so fundamental to AI. Do the current protections around IP, are they fit for purpose, or do we need something different? These are some of the questions that we want to get to in our consultation; are these things fit for purpose, where should adaptations be made, but it also comes back to the balancing the rights of those that own the data with those who want to use it. 

Hayleigh Bosher: So, thinking about future developments, I was reading the UK AI Council Roadmap, where they said that “Government should also consider how to redefine intellectual property to encourage a creative technology ecosystem.” And naturally, I was wondering what does that mean, redefine? Very interesting statement, and there wasn’t very much else on IP in the roadmap. So, I was wondering, if the IPO are working with them and what your thoughts are around their suggestion?

Tim Moss: Obviously the AI council is an independent body and I can’t speak on their behalf as to exactly what they meant. But yes, we do engage with them, and others such as the Office for AI and other colleagues across government, and other offices and WIPO in this space. So, we are engaging hugely around this because, as I said, the more people we engage with the better we can help define the problems and also look at some of the solutions. I mean, I don’t know whether I would use the words ‘redefining intellectual property’, I will leave that to the lawyers, academics and experts to worry about the semantics of what that means. But I think, AI is asking some really big questions around IP, as we’ve discussed, and so I can certainly see that the IP regime has to answer some of these questions and whether that leads to big changes, or not, that is still the big open question. But, where we see our role, and working with others in government, we need to make sure that these legal and regulatory frameworks that we have are fit for purpose and do adapt over time; you can’t just put your head in the sand and say no this is good forever. That is definitely not right, the IP regime has to adapt over time, as it has done historically. AI is asking these questions and I wouldn’t be surprised if changes come around off the back of this, I am just not quite sure what they are at this point in time, because it is so fundamental to what is happening and changing that link around how the current regime has this human invention and creation piece and AI is challenging that. How that adapts is something we want lots of people’s views on. The UK has a great IP regime and AI tech sector, we want to make sure that the two work closely together, so that we can ensure that the IP system supports the AI sector whilst at the same time making sure that we don’t create imbalances for users and others in terms of the IP system. 

Hayleigh Bosher: You’re right, especially from a copyright perspective, it has always evolved to new technologies but what I am hearing around the AI discussions is that we have had to be truly interdisciplinary. For example, at Brunel we now have an AI Research Centre that is multidisciplinary, and we have a course in AI strategy with the idea being that those graduates won’t be creating AI, or regulating AI, they are the strategy person at the top level bringing all these people together. So, related to this, how do you see the IPO’s perspective, in terms of leading the UK IP regime on a global scale?

Tim Moss: The backdrop to this is that IP is both global and national, so you can’t do either one in isolation. The issues around AI are absolutely things we’ve had discussions with a number of Offices around the world, we’ve been involved with conferences with WIPO, we played a key role in pushing that debate forward. I think for IP offices there are two key aspects. One is how we can use AI to improve the IP process and make it more effective. Such as developing AI tools, for trade marks for example we’ve got a customer pre-apply tool which has been really successful, that is using AI to help look at whether a trade mark is likely to be successful in terms of the application, and provide advice and guidance. So, there is all those sorts of things, there are other Offices uses AI around patent applications and patent searching and things like that. So, there is; how do we use AI to make the IP system itself work better and offer a better service to customers. And then we’ve got these really big questions, and it would be an odd world where you have wildly different approaches in different countries. So, ensuring that you’ve got that global debate around these things is hugely important. But, at the same time, also recognising that not all countries are at the same stage in terms of their development of AI. WIPO has 192 members and AI is a big issue for some, but less of an issue for others. And so, it’s about how do you work with those like-minded countries, to say okay how is that developing, which emphasis have they got. The UK has a brilliant creative music sector, more so than many other countries and so it may be that how AI is impacting on music might be an even bigger issue for us than it may be for some other countries. We want to make sure that we can really support the brilliant AI sector that we have in the UK, through the IP regime, but also then make sure that individual creators and inventors also have the necessary protections. There’s also the other side where AI will manage things in terms of infringement, and also some potentially great advantages where AI and algorithms can be used to help stop infringement. For example, some platforms are using AI to identify where there may be counterfeit goods and things like that. It is a fascinating area and one where we absolutely need to engage with the widest possible input to decide whether the IP regime should change, and if so, what should it change to, in order to make sure that it is relevant now and in the future. 

Hayleigh Bosher: Well, thank you so much for your time today and talking to us about this important and interesting area, we will keep an eye out for the forthcoming consultation from the UK IPO on AI and IP.