Habemus a Right to an Explanation: so What? – A Framework on Transparency-Explainability Functionality and Tensions in the EU AI Act

 The European Union’s Artificial Intelligence Act (AI Act),
 finalized in February 2024, mandates transparency and ex
plainability requirements for AI systems to enable effec
tive oversight and safeguard fundamental rights. Yet the
 practical implementation of these requirements faces chal
lenges due to tensions between the need for meaningful ex
planations and their potential risks. This research proposes
 the Transparency-Explainability Functionality and Tensions
 (TEFT) framework to analyze the interplay of legal, tech
nical, and socio-ethical factors shaping the realization of al
gorithmic transparency and explainability in the EU context.
 Through a two-pronged approach combining a focused lit
erature review and an in-depth examination of the AI Act’s
 provisions, we identify key friction points and challenges in
 operationalizing the right to explanation. The TEFT frame
work maps the interests and incentives of various stakehold
ers, including AI providers & deployers, oversight bodies,
 and affected individuals, while considering their goals, ex
pected benefits, risks, possible negative impacts, and context
 to algorithmic explainability.