Trade secrecy in automated decisions: against the myth of irreconcilability and the imposition of patents

Abstract

• Trade secrecy has been widely recognised as an obstacle to transparency in automated decision-making. In multiple cases involving automated decisions impacting individuals, the disclosure of decision-making processes has been restricted to safeguard proprietary information.

• This limitation to transparency has prompted scholars to question the appropriateness of trade secrecy in the age of algorithms including calling for the complete removal of trade secret protection for algorithms in some cases. While some scholars have proposed an unconditional disclosure of the data used in automated decision systems as well as of their source code, others have favoured the use of patents for certain software programmes.

• In this article, based on extensive analysis of existing law including judicial decisions, we discuss how to achieve a balance between these two conflicting positions. Furthermore, we contend that neither abolishing trade secrecy nor mandating patent-based protection for algorithms is necessary or justifiable. Lastly, we assert that, although patenting software and artificial intelligence (AI) is possible, businesses are likely to prefer trade secret protection over patents for their inventions due to the practical challenges of obtaining patents for AI systems.