Trade Rules and Algorithmic Transparency

Regulators all over the world are contemplating various forms of regulation that improve platform accountability, algorithmic transparency and explainability for automated decision-making. For example, AI plays a material role in content moderation, including some context-specific content such as hate speech. When our behavioral data becomes algorithmic input, it is important to empower consumers to understand how a social media algorithm makes content moderation decisions, whether it is appropriate to use that content-moderating algorithm, and how consumers can receive objective algorithmic decisions. Algorithms can be governed by various regulatory choices, including who and what will be regulated, to whom information will be disclosed, and how disclosure will occur. However, a growing number of FTAs contain provisions to restrict access to source code. Several FTAs extend the general prohibition on disclosure requirements to algorithms. In the context of international economic law, the interplay between AI regulation and international trade agreements centers on questions relating to how to ensure that algorithmic accountability is appropriately distributed, as well as how to safeguard public access and oversight over algorithms. More profoundly, how can we balance competing interests in particular trade secrets and prevent AI regulations from being overly trade restrictive?