Following on the heels of the United States Patent and Trademark Office’s request for comments, the World Intellectual Property Organization (WIPO) recently released an issues paper on artificial intelligence (AI) and intellectual property (IP) policy. Comments may be submitted by February 14, 2020.
The WIPO paper identifies AI issues in: (a) Patents, (b) Copyright, (c) Data, (d) Designs, (e) Technology Gap and Capacity Building, and (f) Accountability for IP Administrative Decisions.
In reference to patents, the WIPO paper notes that there are already “several reported cases of applications for patent protection in which the applicant has named an AI application as the inventor,” and asks: (i) should the law permit or require that the AI application be named as the inventor or require that a human be named; (ii) who should be recorded as the owner; and (iii) should the law exclude from patent protection inventions that are generated autonomously by an AI application? The WIPO paper also identifies issues for AI-generated or AI-assisted inventions, including patent eligibility, the status of AI-generated content as prior art, what the standard of a “person skilled in the relevant art” should be, and the disclosure requirements of machine-learning inventions, including whether the data used to train an algorithm should be disclosed or described and what constitutes an adequate disclosure of an algorithm that may change over time. In considering the broader objectives of the patent system in encouraging innovation, the WIPO paper further asks whether consideration should be made to a sui generis system of IP rights for AI-generated inventions.
On copyrights, the WIPO paper raises potential issues with: authorship and ownership of AI applications that produce literary and artistic works; the standard for infringement, and whether the use of data for machine learning constitute an infringement; deep fakes, and to what extent the copyright system should protect deep fakes; and bias in AI applications.
For data, the WIPO paper recognizes that the classic IP system already affords certain types or protection to data, including data that represent inventions, original literary or artistic works, confidential/trade secret material, as well the selection or arrangement of data in jurisdictions that recognize sui generis database rights. The WIPO paper asks whether a new data right should be created, or whether existing IP rights, unfair competition laws, contractual arrangement, and technological measures are sufficient to protect data.
The WIPO paper further discusses authorship and ownership issues in AI-assisted or AI-generated designs, whether there are measures that should be considered in IP policy to reduce the technology gap and AI-capacity gap among countries, and accountability for automated decisions in IP administration and prosecution.
One theme through both the WIPO paper and the USPTO’s request for comments on patenting AI-inventions is whether the existing IP system is sufficient to address the unique challenges posed by existing and contemplated AI-inventions and to foster innovation in this space. Based on current trends, and as noted in the WIPO technology trends report on AI released earlier this year, “AI-related inventions are booming, shifting from theory to commercial applications.” The report found that since AI emerged in the 1950s, nearly 340,000 AI-relation inventions have been filed, with over half of the identified inventions having been published since 2013, and that Japanese, American, and Chinese companies dominate the AI-patenting activity.