Tax & Technology
Tilburg University — TiU II
Open-Book Exam Summary Book
COURSE: Tax & Technology TiU (695526-M-6) · 6 EC · Semester 2
INSTRUCTORS: Prof. dr. A.H. Bomer · R. Lucas MSc · L. Lipiński LLM
EXAM: Thursday, 11 June 2026 · 13:00–16:00 · Open-book (Canvas materials only)
PROHIBITED: Internet · LLMs (ChatGPT etc.) · Google Translate · USB-stick
COVERS: 7 Days: AI · Bias · XAI · ADM · TOM · GDPR · Charter · AI Act · ML · LLMs · Agents · Blockchain
Generated by Perplexity Computer · June 2026 All materials sourced from course Canvas uploads
, Contents
Exam Overview & Tips
Day 1 — AI Fundamentals & Trustworthy AI
– Big Data Process
– Bias (Statistical vs Legal)
– XAI & Trade-off
– ADM
– Profiling
Day 2 — Tax Operating Model
– TOM Components
– Tax Risk Management
– TCF Maturity
– Five Trends
Day 3 — Legal Protection Framework
– Selection Tools
– GDPR Deep Dive
– Charter
– EU Principles
– National Case Law (SyRI, HR)
– AI Act Classification
Day 4 — Machine Learning & Deep Learning
– Traditional vs ML
– Supervised ML
– Model Performance Metrics
– VAT Compliance Case Study
– Model Risks & SHAP
Day 5 — Large Language Models
– LLM Architecture (Attention / Transformers)
– Limitations
– Tax Use Cases
– Generative AI for Tax (Alarie 2024)
Day 6 — AI Agents & Legal Protection
– What is an Agent?
– Tax Agent Pipeline (DAC6 Case)
– LLMs in Practice: Reliability, Confidentiality, Liability
– Duty to Explain — National & EU Rules
Day 7 — Blockchain
– Core Technologies
– Blockchain Types
– Smart Contracts
– Carousel Fraud
– Blockchain vs Alternatives
Key Literature Summary
– Wachter et al. — No Right to Explanation in GDPR
– Lipton — Interpretability of ML
– Alarie — Generative AI for Tax
– Fatz et al. — Blockchain & Tax Compliance
Full Alphabetical Glossary
Tax & Technology TiU — Open-Book Exam Summary Page 2
, EXAM OVERVIEW & STUDY TIPS
Category Detail
Date & Time Thursday, 11 June 2026 · 13:00–16:00
Format Written exam on computer at Tilburg University
Open-book rule Only materials published on Canvas; no USB-stick
Language Dutch or English allowed
Prohibited Internet, LLMs (including ChatGPT), Google Translate
Allowed Blank scrap paper and pens
Scoring Exam score (max 10) + optional 0.5 credit from participant system
What is tested Technology understanding · Legal aspects · Risks & opportunities · Tech
solutions to tax problems
Exam Strategy Tips
• Be concise but always provide supporting arguments — do not just state conclusions.
• Connect technology concepts to legal protection (GDPR, Charter, AI Act) — this is a core exam
theme.
• Distinguish statistical bias from legal bias clearly — the course repeatedly emphasizes this
distinction.
• Know the three-part structure: Big Data process → Frictions → Legal protection layer.
• For ADM questions: always check (1) is it solely automated? (2) does it produce legal effects? (3)
is there a national exception?
• AI Act classification in tax is intentionally ambiguous — the exam likely tests whether you can
argue both sides (high-risk vs recital 59 exclusion).
• For blockchain questions: always ask "do I need all three technologies (P2P + cryptography +
hashing)?" If a shared database suffices, blockchain is not needed.
• LLM questions will likely touch on hallucinations, confidentiality (Samsung), and the
Schwartz/LoDuca case.
Core Exam Themes (recurring across all days)
Theme Key concept
Bias Statistical ≠ Legal; legal bias → model rejection; statistical bias →
sometimes acceptable
Explainability (XAI) Trade-off accuracy vs explainability; transparency + post-hoc
interpretability
ADM Solely automated + legal/significant effects → GDPR Art 22; human
involvement breaks the chain
Profiling GDPR Art 4(4); gathering personal data + automated processing to
place in a group
Tax & Technology TiU — Open-Book Exam Summary Page 3
, Legal protection gaps GDPR (personal data only) + Charter (EU law scope) + National law
→ fragmented
Human oversight Required in all stages; "trolley problem" of accountability; data
literacy essential
AI Act in tax Not clearly high-risk (Recital 59); but probably limited-risk —
transparency obligations apply
Tax & Technology TiU — Open-Book Exam Summary Page 4
Tilburg University — TiU II
Open-Book Exam Summary Book
COURSE: Tax & Technology TiU (695526-M-6) · 6 EC · Semester 2
INSTRUCTORS: Prof. dr. A.H. Bomer · R. Lucas MSc · L. Lipiński LLM
EXAM: Thursday, 11 June 2026 · 13:00–16:00 · Open-book (Canvas materials only)
PROHIBITED: Internet · LLMs (ChatGPT etc.) · Google Translate · USB-stick
COVERS: 7 Days: AI · Bias · XAI · ADM · TOM · GDPR · Charter · AI Act · ML · LLMs · Agents · Blockchain
Generated by Perplexity Computer · June 2026 All materials sourced from course Canvas uploads
, Contents
Exam Overview & Tips
Day 1 — AI Fundamentals & Trustworthy AI
– Big Data Process
– Bias (Statistical vs Legal)
– XAI & Trade-off
– ADM
– Profiling
Day 2 — Tax Operating Model
– TOM Components
– Tax Risk Management
– TCF Maturity
– Five Trends
Day 3 — Legal Protection Framework
– Selection Tools
– GDPR Deep Dive
– Charter
– EU Principles
– National Case Law (SyRI, HR)
– AI Act Classification
Day 4 — Machine Learning & Deep Learning
– Traditional vs ML
– Supervised ML
– Model Performance Metrics
– VAT Compliance Case Study
– Model Risks & SHAP
Day 5 — Large Language Models
– LLM Architecture (Attention / Transformers)
– Limitations
– Tax Use Cases
– Generative AI for Tax (Alarie 2024)
Day 6 — AI Agents & Legal Protection
– What is an Agent?
– Tax Agent Pipeline (DAC6 Case)
– LLMs in Practice: Reliability, Confidentiality, Liability
– Duty to Explain — National & EU Rules
Day 7 — Blockchain
– Core Technologies
– Blockchain Types
– Smart Contracts
– Carousel Fraud
– Blockchain vs Alternatives
Key Literature Summary
– Wachter et al. — No Right to Explanation in GDPR
– Lipton — Interpretability of ML
– Alarie — Generative AI for Tax
– Fatz et al. — Blockchain & Tax Compliance
Full Alphabetical Glossary
Tax & Technology TiU — Open-Book Exam Summary Page 2
, EXAM OVERVIEW & STUDY TIPS
Category Detail
Date & Time Thursday, 11 June 2026 · 13:00–16:00
Format Written exam on computer at Tilburg University
Open-book rule Only materials published on Canvas; no USB-stick
Language Dutch or English allowed
Prohibited Internet, LLMs (including ChatGPT), Google Translate
Allowed Blank scrap paper and pens
Scoring Exam score (max 10) + optional 0.5 credit from participant system
What is tested Technology understanding · Legal aspects · Risks & opportunities · Tech
solutions to tax problems
Exam Strategy Tips
• Be concise but always provide supporting arguments — do not just state conclusions.
• Connect technology concepts to legal protection (GDPR, Charter, AI Act) — this is a core exam
theme.
• Distinguish statistical bias from legal bias clearly — the course repeatedly emphasizes this
distinction.
• Know the three-part structure: Big Data process → Frictions → Legal protection layer.
• For ADM questions: always check (1) is it solely automated? (2) does it produce legal effects? (3)
is there a national exception?
• AI Act classification in tax is intentionally ambiguous — the exam likely tests whether you can
argue both sides (high-risk vs recital 59 exclusion).
• For blockchain questions: always ask "do I need all three technologies (P2P + cryptography +
hashing)?" If a shared database suffices, blockchain is not needed.
• LLM questions will likely touch on hallucinations, confidentiality (Samsung), and the
Schwartz/LoDuca case.
Core Exam Themes (recurring across all days)
Theme Key concept
Bias Statistical ≠ Legal; legal bias → model rejection; statistical bias →
sometimes acceptable
Explainability (XAI) Trade-off accuracy vs explainability; transparency + post-hoc
interpretability
ADM Solely automated + legal/significant effects → GDPR Art 22; human
involvement breaks the chain
Profiling GDPR Art 4(4); gathering personal data + automated processing to
place in a group
Tax & Technology TiU — Open-Book Exam Summary Page 3
, Legal protection gaps GDPR (personal data only) + Charter (EU law scope) + National law
→ fragmented
Human oversight Required in all stages; "trolley problem" of accountability; data
literacy essential
AI Act in tax Not clearly high-risk (Recital 59); but probably limited-risk —
transparency obligations apply
Tax & Technology TiU — Open-Book Exam Summary Page 4
