In its recent study on IP and investment in AI, the UK Intellectual Property Office (“UKIPO”) considered the extent to which the UK’s IP framework influences decisions on UK investment in AI. Ultimately, the study found that the UK’s IP framework is “trusted and reliable and not of concern” while other factors are more critical to investment decisions. We consider some of the study’s key findings below.
The UK’s AI investment landscape
The study states that the UK has earned a prominent reputation as one of the best places in the world to incubate new technologies, with London being the main hub for AI development activity in Europe. It suggests two reasons for this. First, the level of private investment in AI in the UK (primarily by venture capital, private equity and M&A activity) is third globally behind the USA and China and higher than anywhere else in Europe. Second, the UK is one of the leading countries in Europe for the roll out of AI in practice.
Despite this flurry of AI activity, the number of AI-related patent applications made in the UK is relatively low. Germany, France, Japan and South Korea all see more AI-related patent applications than the UK does. And of those applications made in the UK, more are made by US-based applicants than UK-based ones. The study suggests a few reasons for this, including: (i) the limited understanding of the UK’s IP framework within the AI industry; (ii) the draw of the US market; and (iii) the relative importance of other factors driving investment which take priority.
Limited understanding of the UK’s IP framework
There are a number of IP rights that may be available to protect AI (and datasets used for machine learning (“ML”)) in the UK, including copyright, database rights and patents, as well as related rights such as trade secrets. However, the study indicates that organisations’ understanding of these IP rights, and their application to AI, varies depending on their size and sector. For example, small and medium sized enterprises have a better understanding of patent and copyright protection than start-ups but still have a poor understanding of trade secrets. Despite this general trend, interview participants from the technology sector shared a common misconception that “core AI software is not patentable”. In fact, AI software is patentable where it produces a technical effect and the UKIPO grants many thousands of software patents in the field of AI.
The draw of the US market
The study indicates that the relative scale of the US market, and its investment potential, is important too. While many AI businesses are incubated in the UK, the market is seen by many as too small to attract larger investment necessary to scale-up. This means that many of the UK’s AI businesses either opt for an early exit or move to the US (which attracted 50x more private investment in AI than the UK in 2020).
It was also noted that most AI start-ups tend to rely on copyright, trade secrets and confidentiality obligations to protect their AI. This does not align with many investors’ criteria. Many (rightly or wrongly) require organisations to have patents, or to have made patent applications, before they decide to invest. According to the study, this is because patents are seen by investors as an endorsement of the AI by the granting authority and as saleable assets in the event the organisation stops trading. The study goes on to explain that as many AI business in the UK are in their early stages, their AI may not be sufficiently developed to obtain patent protection. However, once they receive investment to scale-up their activities, patenting may become achievable. Investment from US investors, or investors that want the AI business to enter the larger US market, may explain the study’s finding that UK companies choose to file patents in the US instead of the UK.
Other key investment factors
The study identified a number of other factors that dominate decisions to invest in AI in the UK: (i) the availability of AI-skilled people; (ii) the reputation of British technology hubs and universities; (iii) UK government grants and access to early-stage funding; and (iv) the fact that the UK is naturally English speaking in common with the USA. In light of these key factors, the study concludes that the UK’s IP framework is seen as a “background hygiene factor” and not a pivotal consideration for investors.
It is clear that AI is high up on the UK Government’s agenda. This year alone the UK Government has: (i) launched the UK’s AI standards initiative, which aims to shape global technical standards for AI (see our blog here); (ii) published its UK Digital Strategy, which promises a "light-touch, pro-growth regulatory regime" (see our blog here); and (iii) published its AI policy paper (which outlines its (pro-innovation) approach to regulating AI in the UK) and AI Action Plan (which sets out how it will deliver against the National AI Strategy) - see our blog here.
Whilst this study found that the UK’s IP framework is not a fundamental consideration for investment in AI for the participants involved, IP clearly still has an important role to play in AI. Only a few days before this study was published, the UKIPO announced, in response to its recent consultation on AI and IP, that it will introduce an exception to existing copyright and database rules to allow text and data mining for any purpose, including for training AI systems and supporting the adoption and development of ML (see our blog here).
What effect these measures will have on UK investment in AI remains to be seen but we will continue to watch these developments closely.