Implementing the Transparency Chapter of the Code of Practice on General-Purpose AI Models: A Customized Practical Manual for Startups

Abstract

This thesis examines how AI startups that provide general-purpose AI models can operationalise the transparency obligations laid down in Article 53(1)(a) and (b) of the EU AI Act in a practical and value-creating way. The analysis focuses on the interplay between the AI Act’s binding requirements and the voluntary soft-law instrument of the General-Purpose AI Code of Practice, with particular attention to its transparency chapter. This thesis demonstrates how the Code of Practice can translate legal obligations into concrete measures and structured documentation, and how this can function as a tool for the obligations towards the EU AI Office and competent authorities as well as downstream business customers. It further explores governance and compliance strategies for startups, including how internal decision-making and corporate governance structures can be used to make Code-based commitments effectively binding within the organisation and embedded into product development and business processes.