LSTS and HALL researcher Anastasiya Kiseleva published her new blogpost 'MAKING AI’S TRANSPARENCY TRANSPARENT: notes on the EU Proposal for the AI Act' on the European Law Blog.
She explores how transparency is reflected in the proposed AI Act and how it correlates with the related concepts: 1) communication, 2) interpretability, 3) explainability, 4) information provision, and 5) record keeping and documentation. Her analysis demonstrates that the EU legislator should have a more coherent vision on the ‘transparency’ terminology in the AI Act. For consistency, she suggests establishing the hierarchy between the related concepts where transparency shall be seen as the broadest one. In a legal sense, it can be seen as the principle of the AI Act. This way, the other relevant elements can be seen as the measures to ensure the principle. This hierarchy would follow the approach already existing in the legislation (such as the GDPR).
You can read the full blogpost on the European Law Blog.