Lawyers usually do well on AIGP Domain II (laws and frameworks) but lose points in Domains I, III and IV: AI concepts and lifecycle governance, which together make up about three-quarters of the exam. The fix is to spend most of your study time on AI fundamentals and operational scenarios, and to answer like an AI governance lead, not outside counsel. Choose the practical, correctly sequenced governance action rather than the most legally complete analysis.
Legal professionals bring real strengths to the AIGP: reading regulations precisely, understanding roles and liability, and comfort with ambiguity. But the exam is designed for AI governance practitioners, and some legal habits work against you. Here's how to adjust.
Where legal training helps
- Domain II: the EU AI Act's structure, privacy law, liability, IP and anti-discrimination law.
- Roles and accountability: provider versus deployer maps naturally onto legal thinking about responsibility.
- Close reading of questions, which helps with qualifiers and multi-select items.
Where legal training hurts
| Legal habit | Why it costs points | Fix |
|---|---|---|
| Looking for the legally complete answer | The exam often wants the practical first step | Ask: "What would the governance lead do next?" |
| Over-reading facts into scenarios | Adding assumptions changes the answer | Answer only on the facts given |
| Treating everything as a compliance question | Many questions are about engineering practice, testing or monitoring | Learn the AI lifecycle and its controls |
| Skipping AI fundamentals | Domain I terms (drift, fine-tuning, RAG, evaluation) appear across the whole exam | Spend a full week on AI concepts |
| Over-investing in the EU AI Act | AI-specific law is only about 6–8 scored questions | Cap Act study; practise scenarios instead |
The AI concepts lawyers most often need to learn
- Supervised, unsupervised and reinforcement learning; generative versus predictive models.
- Training, validation and test data; overfitting; data drift and concept drift.
- Foundation models, fine-tuning, retrieval-augmented generation (RAG), prompts and system prompts.
- Evaluation metrics (accuracy, precision and recall at a conceptual level) and fairness metrics.
- Explainability techniques at a high level, and why some models are opaque.
- AI agents and tool use.
A 6-week plan for legal professionals (about 50 hours)
- Week 1 (10 h): AI fundamentals and the lifecycle, with no law at all. Take a diagnostic quiz.
- Week 2 (6 h): Domain II revision: EU AI Act roles, tiers and timeline; other AI laws; frameworks. Keep it efficient.
- Week 3 (9 h): Domain III, governing development: data governance, testing, documentation, impact assessments.
- Week 4 (9 h): Domain IV, governing deployment: procurement, oversight, monitoring, incidents, agentic AI.
- Week 5 (9 h): Mixed scenario practice, focused on Domains I, III and IV.
- Week 6 (7 h): Full timed mock, review and cram.
Career angle
For lawyers, the AIGP signals that you can advise on AI products and governance, not only on regulation. Pairing it with a CIPP strengthens privacy credibility. See AIGP vs CIPP and AI governance jobs in 2026.
Practice: choose the governance answer (original Archuz item)
A product team wants to add a generative AI summarisation feature to a customer support tool next month. As the AI governance lead with a legal background, what should you do first?
- Draft a 20-page legal memo on every applicable law
- Run a proportionate intake: identify the data involved, who sees outputs, and the risks; check transparency, privacy and vendor terms; agree the testing needed before launch
- Refuse until all regulations are final
- Approve because summarisation is low-risk
Answer: B. It's proportionate, action-oriented and correctly sequenced. A is thorough but not the first step, C is disproportionate, and D skips the assessment.
Turning legal skills into governance answers
| Legal instinct | Governance answer the exam prefers |
|---|---|
| "It depends on the jurisdiction." | Apply the most relevant law in the scenario and choose the proportionate control |
| "Get a legal opinion first." | Involve legal, but also run the assessment, testing and documentation |
| "Add a contract clause." | Contracts help, but pair them with due diligence, testing and monitoring |
| "Read the statute carefully." | Know the framework, then decide what happens next at this lifecycle stage |
Common legal practice scenarios on the AIGP
- Classifying a system under the EU AI Act and identifying the provider and deployer.
- Deciding whether GDPR Article 22 applies to an AI-supported decision.
- Allocating responsibilities in a vendor contract for an AI system.
- Assessing copyright and licensing risks in training data.
- Advising on disclosures for chatbots and AI-generated content.
Two practice questions for lawyers
Original Archuz questions, not real exam items.
1. A client asks whether its AI hiring tool is "legal". What is the best first governance response? A. Give a yes or no opinion. B. Classify the use (high-risk under the EU AI Act, a likely DPIA under the GDPR, local hiring laws), then identify the required assessments, controls and documentation. C. Recommend not using AI. D. Rely on the vendor's assurances. Answer: B.
2. Which contract term best addresses the risk that a vendor changes its model without notice? A. A confidentiality clause. B. A change notification obligation with the right to re-assess or terminate. C. A longer contract term. D. A non-solicitation clause. Answer: B.
Glossary lawyers should master
- Training vs inference: building the model vs using it to produce outputs.
- Fine-tuning: further training of a pre-trained model on specific data.
- RAG: retrieval-augmented generation, where the model uses retrieved documents to answer.
- Drift: performance changes as real-world data diverges from training data.
- Hallucination or confabulation: plausible but false outputs.
- Agent: an AI system that plans and takes actions using tools.
Close the gap where lawyers lose points
The AIGP Complete Pack's domain-tagged questions focus your practice on Domains I, III and IV, where legal training helps least.
Get the Complete PackFrequently asked questions
Is the AIGP hard for lawyers?
Lawyers usually find Domain II manageable but need to work hard on AI concepts and lifecycle governance, which make up most of the exam.
Does the AIGP count for CLE credits?
The exam itself isn't CLE. Some AIGP training courses may be CLE-accredited depending on your jurisdiction. Check with your bar.
Should lawyers take the CIPP or AIGP first?
If privacy law is new to you, the CIPP first. If you already advise on privacy, the AIGP adds AI-specific credibility.
What is the biggest adjustment lawyers need to make for the AIGP?
Moving from legal analysis to governance decisions: choosing proportionate, practical actions at the right lifecycle stage rather than seeking a definitive legal interpretation.
How do lawyers usually perform on AIGP law questions?
Well on legal concepts, but they can lose points by over-analysing or choosing legal opinions over practical governance actions. Practising scenario questions helps.
Sources
Last reviewed: October 2026.