'Trade Secrecy, Factual Secrecy and the Hype Surrounding AI' by Sharon K. Sandeen and Tanya Aplin, in Ryan Abott (ed) Research Handbook on Intellectual Property and Artificial Intelligence (Edward Elgar, forthcoming) comments
Access to and sharing of anonymized machine-generated data and the transparency of data analysis techniques has taken on vital importance in a world characterized by artificial intelligence, particularly machine learning'. In short, this chapter interrogates the extent to which such data and algorithms may qualify as 'trade secrets' under US and EU trade secrets law, focusing in particular on whether the definition of a ‘trade secret’ is met. We show through the use of two case studies – involving autonomous vehicles and credit scoring – and a close analysis of the trade secrets definition that the claim of trade secrets protection is overstated. The greater risk relates to factual secrecy rather than legally protected trade secrets and the policy debate needs to shift to assess what regulation, if any, there should be of data that is simply kept secret.
