Abstract

The chapter examines whether the Council of Europe’s Framework Convention on Artificial Intelligence and Human Rights envisages a requirement for human oversight of AI systems. Although the Convention’s final text does not explicitly impose such a duty, human oversight featured prominently throughout the drafting process and remains referenced in the Explanatory Report. After retracing this drafting history, the chapter evaluates the interpretative value of the Explanatory Report under Articles 31 and 32 of the Vienna Convention on the Law of Treaties. It argues that while the Report cannot be regarded as an authoritative source of legal obligations, its reference to human oversight offers important interpretive guidance. Yet, it is contended that a requirement for human oversight can be derived from the broader human rights framework the Convention is meant to operationalize, particularly the doctrine of positive obligations under human rights law. Finally, the article outlines practical conditions for ensuring that human oversight functions as an effective safeguard, emphasizing design, competence, organizational context.

TABLE OF CONTENTS

1. Introduction

2. The human oversight requirement in the drafting history of the AI Framework Convention

3. The legal value of the reference to human oversight in the Explanatory Report

4. Human Oversight as a Positive Obligation under Human Rights Law

5. How to Ensure that Human Oversight Effectively Serves as a Safeguard?