Transparent AI? Navigating Between Rules on Trade Secrets and Access to Information

Abstract

AI systems are nowadays employed in ever-increasing areas. This new era of technological development is exciting, but AI applications are also a cause for concern. If tasks that have hitherto normally been undertaken by human beings are now to be taken care of by ever more intelligent autonomous systems, how can we be certain that such functions are performed diligently and safely? Many areas of application of AI systems have also made the tribulations of AI utilization apparent. The EU’s Artificial Intelligence Act (AIA) aims to tackle the concerns and challenges related to the utilization of AI, and to develop human-centric, secure, trustworthy, and ethical AI systems for the EU markets. The provisions of the AIA establish a system of compliance assessment that requires AI providers to disclose how high-risk AI systems have been trained and put together. This article will look at the role of disclosure obligations under the provisions of the AIA. The focus is on the tension between obligations to disclose information on the one hand and requirements to protect the trade secrets contained in the technical details of AI on the other. This article will explain how the technical details of AI contain some information that does not qualify for trade secret protection. And even when there are trade secrets, there are exceptions to trade secret protection. Rules to enable access to information form part of the Trade Secrets Directive, but other legislative instruments too enable access and make it necessary to navigate between access and confidentiality.