Disclosing Without Divulging: Trade Secrets in the EU AI Act’s Transparency Duties

Abstract The transparency obligations of the EU Artificial Intelligence Act and the trade secret protections of the Trade Secrets Directive are often portrayed as standing in fundamental, zero-sum tension. This article challenges that portrayal: while individual disclosure decisions are necessarily binary in form, the regime-level relationship between the two instruments admits of calibrated reconciliation. Introducing a three-vector typology that distinguishes authority-facing, supply-chain, and individual-facing transparency, it offers a novel doctrinal analysis of the conflict surface between the two instruments structured around these distinct disclosure audiences. The analysis yields three findings. First, the cumulative requirements of Art. 2(1) TSD and the functional explanation standard from CK v. Dun & Bradstreet Austria (C-203/22) meaningfully narrow the zone of genuine conflict. Second, Art. 78 of the AI Act establishes a five-ground confidentiality architecture that extends beyond trade secret doctrine, rendering TSD-focused reform proposals partial. Third, existing doctrinal tools, including TSD Art. 3(2), the scope-level deferral in Art. 1(2)(b), and the Varec procedural model, provide reconciliation pathways for authority-facing and supply-chain vectors. The critical gap lies not in the substantive law but in the absence of a dedicated procedural pathway for individual-facing Art. 86 disputes.