UK strives to “be a leader in AI technology” as it sets out its next steps for the regulation of artificial intelligence.

By Deborah J. Kirk, Laura Holden, and Victoria Wan

On 23 March 2021, the UK Intellectual Property Office (IPO) published the outcome of its consultation last year on artificial intelligence (AI) and intellectual property (IP) (the Response). The Response highlights the UK’s ambition to “be a leader in AI technology” by developing and adapting IP legislation in light of developments in AI technology, notwithstanding the fact that such developments may, post-Brexit, result in a divergence between the approach taken in the UK and the EU.

Since the UK’s withdrawal from the EU, the UK has clearly indicated that it will take its own path to encourage AI innovation while protecting IP rights to solidify its position as a leader in AI innovation. The UK has been active in its attempt to secure this position. The Response follows the AI Council’s publication of an AI Roadmap in January this year and the House of Lords Liaison Committee’s publication of AI in the UK: No Room for Complacency in December last year, two reports that recognise the importance of good governance and regulation for public trust while specifying that flexible regulation is critical. The Response is an example of one of the actions recommended by the House of Lords’ Liaison Committee for sector-specific regulators (such as the IPO) to identify gaps in regulation (such as IP legislation) to address issues raised by AI.

Key Findings

The Response summarises the views of more than 90 rights holders, developers, and users of AI technology, from large organisations to individuals. The key findings are as follows:

  • Respondents expressed concerns that the inventorship criteria of securing a patent may impact patent availability for AI-generated inventions, which could result in a reduced incentive to invest in AI research. Respondents also emphasised the need for clarity and predictability in the IPO’s patent exclusion practices in relation to AI technology. The majority of respondents agreed that patents play an important role in protecting and supporting AI innovation, but there were mixed views on the extent to which patents should be made available to protect these technologies.
  • Respondents indicated that human creators should be “put first” over works created by AI technology. Some considered that AI generated works should not be protected by copyright at all and instead potentially benefit from a separate right with lesser duration and scope, reflecting the investment in such works. Respondents also flagged the importance of text and data mining (TDM), a process involving selecting, accessing, copying, and analysing large amounts of copyright material, in training AI technology and queried whether obtaining licences granting permission to use such copyright material was adequate and sufficiently available to ensure that TDM does not constitute an infringement. Some respondents queried whether the copyright exception allowing TDM for the purposes of non-commercial scientific research should be expanded to include commercial uses. Rights holders felt that voluntary licensing of works used in TDM would be preferential, since expanding the copyright exception could shift the balance unfairly against creators.
  • Respondents generally felt that the law in the areas of trademarks, designs, and trade secrets was adequate and flexible enough to respond to the present and immediate challenges presented by AI technology but that the law may need to be revisited as AI technology develops. In particular, if AI systems were to advance to a stage where they could operate independently, without apparent control, then it could be difficult to determine where liability for infringement of trademarks should fall (e.g., if an AI system operates a website that offers consumers infringing products, it may be difficult to determine whether liability for infringement is with the seller, the AI enabling the sale, or the ultimate recipient). At present, human influence remains predominant for AI systems, and respondents felt that AI technology is simply a tool and possible medium for infringement and not an infringer.

Next Steps

The Response makes clear that the UK aims to become “the best place in the world for research and innovation” and to be “at the forefront of the artificial intelligence and data revolution”. In order to achieve this, the Response sets out proposed next steps to encourage innovation in AI while preserving the role of IP in promoting human creativity and innovation. The key steps identified in the Response are as follows:

  • Conduct Research. The IPO plans to (i) research AI and IP enforcement, including whether the UK’s copyright enforcement regime is adequate to allow copyright owners to act against large-scale copyright infringement by AI developers who use copyright-protected works without authorisation to train AI (with findings to be published this fall); (ii) commission an economic study into the role of IP in incentivising investment in the AI sector; and (iii) undertake a feasibility study of a “deposit” system to file AI training data disclosed within patent applications at the IPO.”
  • Consult with Stakeholders. The IPO will continue to consult on AI and in particular:
  • Patents: The IPO will call for views later this year on the range of possible patent policy options, including legislative change, for protecting AI-generated inventions that would not otherwise meet inventorship criteria.
  • Copyright: The IPO proposes to consult on whether to limit copyright protection to human creation only, preventing AI-generated works from benefiting from copyright protection, and instead protecting AI with a related right with a lesser scope and investment. The IPO will also review the ways in which copyright owners licence their works for AI training purposes and consult on measures to make this easier.
  • Update Guidelines. The IPO plans to publish enhanced guidelines on patent exclusion practices for AI inventions. As part of this, the IPO will review its patent practices and establish any differences in outcome for AI patent applications filed at the IPO vs. the European Patent Office (EPO) (as some respondents flagged that the EPO’s more permissive patent exclusion approach gave a better outcome for AI patent applications compared to the IPO). Whether this comparison shows an intention to align with the EPO’s arguably more permissive approach or further diverge from the EPO approach remains to be seen.
  • Build Reputation. The IPO will continue to engage with like-minded nations and organisations to deepen understanding and foster cooperation with the intention of leading the global debate on policy approaches that balance growth with IP protection. As part of this, the IPO will hold UK-wide seminars starting this spring that build on the content of the Response, starting with a joint seminar with the Alan Turing Institute (the UK’s national research centre for AI). See here and here for the UK government’s and the ICO’s other recent collaborations with the Alan Turing Institute.
  • The IPO will monitor developments in AI technology and keep issues in regards to trademarks, designs, and trade secrets under review. The IPO does not consider urgent action in these areas to be required.

Going Forward

Organisations should expect rapid development in the relationship between AI technology and IP this year, including a joint seminar hosted by the IPO and the Alan Turing Institute this spring and the IPO’s publication of its report on AI and IP enforcement this autumn. More widely, other sector-specific regulators will likely continue to identify gaps in their respective sectors’ regulations and publish reports or calls for views.

Latham & Watkins will continue to provide updates on the regulation of AI in the UK.