Movie buffs may recall HAL 9000 as the fictional AI character and main antagonist in Stanley Kubrick’s and Arthur Clarke's Space Odyssey series. Although the film recently celebrated its 50 year anniversary, with rapid recent advances in AI technology, there are still many questions in respect of IP rights for AI machines like HAL. AI-based businesses will be interested to know that the World Intellectual Property Office (WIPO) is continuing its policy work on the intersection of AI and IP rights.

As discussed in our earlier blog posts, recent patent office decisions by the EPO, UK IPO and USPTO highlight some of the theoretical issues around whether owners of AI machines should be able to obtain patent protection for inventions that do not meet human-based criteria for inventorship. Issues of ownership and originality have also been raised in the context of copyright law where ‘robotic artists’ are involved.  As AI increases in importance, WIPO has identified such issues as among the most pressing questions facing IP policymakers.  On 29 May 2020, WIPO published a revised issues paper as part of a broader series of consultations on the intersection between IP and AI. The paper reflects more than 250 submissions received from Member States, academic, scientific and private organizations as well as individuals following the first WIPO Conversation on AI and IP, convened in September 2019.

The substantive legal issues collated by WIPO cover the full gamut of IP rights, including patents, copyright, data, designs, trade marks, trade secrets and issues of unfair competition and can be grouped into three main themes:

Protection of IP rights in connection with AI-generated and AI-assisted outputs

  • Should the law distinguish between AI-generated and AI-assisted outputs (the former being generated without human intervention at all)? If so, what level of human intervention is considered material? 
  • To what extent should an inventor/creator/designer be required to be a human person? Should an AI application be permitted to be named as the inventor on a patent application or the owner of a copyright work generated by that application? What about new AI-generated designs with individual character?
  • To what extent should AI-generated outputs benefit from patent, copyright and design rights? 
  • Is an entirely new sui generis system of IP rights required to re-balance innovation incentives for AI?
  • Is there a need to create new rights in data? Should data and AI applications be protectable as trade secrets? If yes, what exceptions may be required? 

Use of AI as a tool to assist in the administration and management of IP rights

  • To what extent should Member States be permitted to rely on AI when making decisions relating to the prosecution and administration of IP? For example, to conduct prior art searches when prosecuting a patent application.
  • Are legislative or regulatory changes desirable to facilitate or address the consequences of decision-making by AI applications? What type of appeal mechanisms are required to deal with decisions made by AI?

Impact of IP on the transparency of AI applications 

  • If AI applications were excluded from patentability would that incentivise related applications and algorithms to be kept secret, thereby increasing issues of transparency (the black box problem)?
  • Is copyright an appropriate vehicle for the regulation of "deep fakes" whereby works are created to simulate the likenesses of a person and their attributes? Should the deep fake generated from underlying copyright material also benefit from copyright protection and if so, who should own the copyright? 
  • Should there be a system of equitable remuneration for persons whose likeness and “performance” are used in a deep fake?

The paper also highlights concerns around a so-called "technology gap" where a small group of countries with existing AI expertise and capacity may advance at an increasingly rapid pace, leaving others behind.

There are no easy answers to the questions raised in the issues paper - however WIPO hopes to advance a more structured dialogue on these issues by bringing together the various strands of discussions that are already taking place within IP offices and organisations, including the European Parliament.  We expect these issues to generate significant debate at the Second Session of the WIPO Conversation on IP and AI which will take place virtually over three days in early July 2020.