AI ETHICS AND REGULATION
Comprehensive Course Text
KU Leuven CiTiP | Academic Year 2025–2026
Prof. dr. Aleksandra Kuczerawy | Prof. dr. Jan de Bruyne & guest lecturers
,Course Overview
The course 'AI Ethics and Regulation' (AIER) is offered by the Centre for IT & IP Law
(CiTiP) at KU Leuven. It addresses the legal and ethical dimensions of artificial
intelligence, with particular focus on the European regulatory framework.
Course Objectives
• Acquire a sound understanding of the legal and ethical issues surrounding AI
• Understand the interaction between ethics and law
• Be able to apply normative principles in risk-anticipation processes and mitigation
strategies
• Critically reflect on AI innovations
Assessment
• Written exam: 16/20 — closed book, 2 hours, open/essay questions
• Group assignment: 4/20 — presentation in groups of 4–5 students on an
assigned statement
Schedule
• 12/2 — Introduction to the course (Kuczerawy)
• 19/2 — Relevant frameworks: AI Act, CAI, AI HLEG (Kuczerawy)
• 26/2 — AI Ethics (dr. Anastasia Siapka)
• 5/3 — AI and liability (Prof. Jan De Bruyne)
• 12/3 — AI and data protection (dr. Pierre Dewitte)
• 19/3 — AI and explainability / XAI (dr. Maja Nisevic)
• 26/3 — AI and law enforcement (dr. Irmak Erdogan)
• 2/4 — AI and intellectual property (dr. Kasper Drazewski)
• 23/4 — AI and standardisation (Koen Vranckaert)
• 30/4 — AI and public administration (Prof. Nathalie Smuha)
• 7/5 — AI and media law (Dutkiewicz & Kuczerawy)
• 21/5 — AI, discrimination and ecological sustainability
,Lecture 1: Introduction — Regulating AI and
Foundational Concepts
1.1 What Does 'Regulating' AI Mean?
Regulating AI goes beyond 'the law' alone. It concerns steering, governing, and
influencing human behaviour and practices, sometimes by opening or closing certain
options for action. Ideally, regulation steers actions and practices in a responsible and
adequate manner.
Regulation has two core elements: (1) setting norms, and (2) enforcing and/or ensuring
compliance with those norms.
Key idea: 'Regulation' = norm-setting + enforcement/compliance. This is broader
than 'the law' alone.
1.2 Law: Basic Concepts
Legal subjects and the hierarchy of norms
Law operates through legal subjects (entities with rights and obligations). There is an
important distinction between 'law in books' (law as written) and 'law in action' (law as
applied in practice).
The legal hierarchy in the EU comprises: primary EU law (Treaties), secondary EU law
(Regulations, Directives), and national law. Regulations are directly applicable in all
Member States; Directives must be transposed into national law.
Principles of EU Law
• Principle of subsidiarity: the EU acts only if the objectives cannot be sufficiently
achieved by the Member States
• Principle of conferral: the EU has only those competences conferred upon it by
the Treaties
• Principle of proportionality: EU action shall not exceed what is necessary to
achieve the objectives of the Treaties
Law ≠ Ethics
Law and ethics are closely related but not identical. Not all law is ethical, and not
everything ethical is legally required. This distinction is crucial for the study of AI
regulation.
1.3 Modalities of Regulation (Lessig)
Lawrence Lessig identifies four modalities of regulation, each with its own advantages
and disadvantages:
1. The Law
• Advantages: broad scope of application, democratic legitimacy, enforceability
• Disadvantages: time-consuming, often lacks expertise, global framework not
realistic, sometimes a blunt instrument
, Function: 'command and control' — the law prohibits, commands or regulates
behaviour.
2. The Market
Adam Smith's 'invisible hand': best results through free interplay of supply and demand.
Advantages: efficiency. Disadvantages: lack of transparency (especially for high tech),
short-term perspective, conditioned preferences (Galbraith).
3. Technology ('Code as Law')
Rules are built into technology. Examples: Digital Rights Management, privacy by
design. Advantages: efficiency and adaptiveness. Disadvantages: when delegated to
private parties — legitimacy risks; large-scale introduction may cause society to forget
underlying norms.
4. Social Norms / Ethical Framework
Motivation, easier enforcement, adaptable. Disadvantages: possible partiality regarding
own and other stakeholders' interests; how to create moral awareness?
📌 Exam tip: Explain the four modalities of Lessig and illustrate each with an example from
an AI context.
1.4 Technology and Society
Four theoretical approaches
• Technological determinism: technological evolutions are autonomous, neutral
and determining for human behaviour (Veblen, Ellul, Postman). Examples:
smartphones, Gen-AI
• Technological instrumentalism: technology is merely an instrument; ethical
considerations arise in its application. The user determines whether use is 'good'
or 'bad'
• Technological substantivism: technology is not neutral (has its own values, cf.
bias); the nature of the technology is determining, not the use
• Critical theory (Feenberg): technology is value-driven, influenced by people who
are themselves influenced by social pressures and their own values. Autonomy of
the user plays a role
Summary matrix: Determinism (autonomous + neutral) → Instrumentalism (human-
controlled + neutral) → Substantivism (autonomous + value-driven) → Critical theory
(human-controlled + value-driven).
Comprehensive Course Text
KU Leuven CiTiP | Academic Year 2025–2026
Prof. dr. Aleksandra Kuczerawy | Prof. dr. Jan de Bruyne & guest lecturers
,Course Overview
The course 'AI Ethics and Regulation' (AIER) is offered by the Centre for IT & IP Law
(CiTiP) at KU Leuven. It addresses the legal and ethical dimensions of artificial
intelligence, with particular focus on the European regulatory framework.
Course Objectives
• Acquire a sound understanding of the legal and ethical issues surrounding AI
• Understand the interaction between ethics and law
• Be able to apply normative principles in risk-anticipation processes and mitigation
strategies
• Critically reflect on AI innovations
Assessment
• Written exam: 16/20 — closed book, 2 hours, open/essay questions
• Group assignment: 4/20 — presentation in groups of 4–5 students on an
assigned statement
Schedule
• 12/2 — Introduction to the course (Kuczerawy)
• 19/2 — Relevant frameworks: AI Act, CAI, AI HLEG (Kuczerawy)
• 26/2 — AI Ethics (dr. Anastasia Siapka)
• 5/3 — AI and liability (Prof. Jan De Bruyne)
• 12/3 — AI and data protection (dr. Pierre Dewitte)
• 19/3 — AI and explainability / XAI (dr. Maja Nisevic)
• 26/3 — AI and law enforcement (dr. Irmak Erdogan)
• 2/4 — AI and intellectual property (dr. Kasper Drazewski)
• 23/4 — AI and standardisation (Koen Vranckaert)
• 30/4 — AI and public administration (Prof. Nathalie Smuha)
• 7/5 — AI and media law (Dutkiewicz & Kuczerawy)
• 21/5 — AI, discrimination and ecological sustainability
,Lecture 1: Introduction — Regulating AI and
Foundational Concepts
1.1 What Does 'Regulating' AI Mean?
Regulating AI goes beyond 'the law' alone. It concerns steering, governing, and
influencing human behaviour and practices, sometimes by opening or closing certain
options for action. Ideally, regulation steers actions and practices in a responsible and
adequate manner.
Regulation has two core elements: (1) setting norms, and (2) enforcing and/or ensuring
compliance with those norms.
Key idea: 'Regulation' = norm-setting + enforcement/compliance. This is broader
than 'the law' alone.
1.2 Law: Basic Concepts
Legal subjects and the hierarchy of norms
Law operates through legal subjects (entities with rights and obligations). There is an
important distinction between 'law in books' (law as written) and 'law in action' (law as
applied in practice).
The legal hierarchy in the EU comprises: primary EU law (Treaties), secondary EU law
(Regulations, Directives), and national law. Regulations are directly applicable in all
Member States; Directives must be transposed into national law.
Principles of EU Law
• Principle of subsidiarity: the EU acts only if the objectives cannot be sufficiently
achieved by the Member States
• Principle of conferral: the EU has only those competences conferred upon it by
the Treaties
• Principle of proportionality: EU action shall not exceed what is necessary to
achieve the objectives of the Treaties
Law ≠ Ethics
Law and ethics are closely related but not identical. Not all law is ethical, and not
everything ethical is legally required. This distinction is crucial for the study of AI
regulation.
1.3 Modalities of Regulation (Lessig)
Lawrence Lessig identifies four modalities of regulation, each with its own advantages
and disadvantages:
1. The Law
• Advantages: broad scope of application, democratic legitimacy, enforceability
• Disadvantages: time-consuming, often lacks expertise, global framework not
realistic, sometimes a blunt instrument
, Function: 'command and control' — the law prohibits, commands or regulates
behaviour.
2. The Market
Adam Smith's 'invisible hand': best results through free interplay of supply and demand.
Advantages: efficiency. Disadvantages: lack of transparency (especially for high tech),
short-term perspective, conditioned preferences (Galbraith).
3. Technology ('Code as Law')
Rules are built into technology. Examples: Digital Rights Management, privacy by
design. Advantages: efficiency and adaptiveness. Disadvantages: when delegated to
private parties — legitimacy risks; large-scale introduction may cause society to forget
underlying norms.
4. Social Norms / Ethical Framework
Motivation, easier enforcement, adaptable. Disadvantages: possible partiality regarding
own and other stakeholders' interests; how to create moral awareness?
📌 Exam tip: Explain the four modalities of Lessig and illustrate each with an example from
an AI context.
1.4 Technology and Society
Four theoretical approaches
• Technological determinism: technological evolutions are autonomous, neutral
and determining for human behaviour (Veblen, Ellul, Postman). Examples:
smartphones, Gen-AI
• Technological instrumentalism: technology is merely an instrument; ethical
considerations arise in its application. The user determines whether use is 'good'
or 'bad'
• Technological substantivism: technology is not neutral (has its own values, cf.
bias); the nature of the technology is determining, not the use
• Critical theory (Feenberg): technology is value-driven, influenced by people who
are themselves influenced by social pressures and their own values. Autonomy of
the user plays a role
Summary matrix: Determinism (autonomous + neutral) → Instrumentalism (human-
controlled + neutral) → Substantivism (autonomous + value-driven) → Critical theory
(human-controlled + value-driven).
