AIGP Glossary
The terms you will meet on the exam and in the regulation, in plain English.
How to use this. Definitions are simplified for beginners. Where a term is legal or technical, check the exact wording in the source text before you rely on it.
- AIGP
- AI Governance Professional, the IAPP certification this site prepares you for.
- IAPP
- The International Association of Privacy Professionals, which runs the AIGP exam.
- Body of Knowledge (BoK)
- The IAPP's published outline of what the exam covers, with four domains and question ranges.
- EU AI Act
- Regulation (EU) 2024/1689, the EU's risk-based law on artificial intelligence.
- Provider
- In the AI Act, the party that develops a system or places it on the market under its name. Check the Article 3 definition.
- Deployer
- In the AI Act, the party that uses an AI system under its own authority. Check the definition in Article 3.
- High-risk system
- A category with specific obligations under the Act. Article 6 and Annex III set out how systems are classified.
- Prohibited practice
- A use the Act bans, listed in Article 5, subject to limited exceptions.
- General-purpose AI (GPAI) model
- A model that can perform a wide range of tasks. The Act has separate rules for these.
- FRIA
- Fundamental rights impact assessment. Certain deployers of certain high-risk systems must complete one (Article 27).
- DPIA
- Data protection impact assessment, required in some cases under GDPR Article 35.
- NIST AI RMF
- The US voluntary AI risk framework, organised around Govern, Map, Measure and Manage.
- ISO/IEC 42001
- An international standard for an AI management system that organisations can be certified against.
- ISO/IEC 42005
- A standard about AI system impact assessment. Check its scope before relying on it.
- Data drift
- A change in the distribution of the inputs a model receives over time.
- Concept drift
- A change in the relationship between inputs and the outcome the model predicts.
- Provenance
- Where data came from, who owns it and under what licence it may be used.
- Lineage
- The record of how data was transformed and moved through a pipeline over time.
- Interpretability
- How well a person can understand the internal workings of a model.
- Explainability
- Giving reasons for a specific output. Usage varies between sources.
- Automation bias
- Over-reliance on an automated output, so the human reviewer stops checking it.
- Human oversight
- People with the knowledge, time and authority to understand, question and override a system.
- TEVV
- Testing, evaluation, verification and validation. Definitions vary, so use your course's wording.
- Membership inference
- An attack that tries to work out whether a particular record was used to train a model.
- Prompt injection
- Manipulating an AI system through crafted inputs, such as hidden instructions in a web page.
- Agentic AI
- Systems that take multi-step actions, often by using tools, rather than only producing text.
- Least privilege
- Giving a system or person only the access needed for the task.
- Shadow AI
- AI use inside an organisation that is not approved, tracked or governed.
- Model card / datasheet
- Documentation of a model (model card) or a dataset (datasheet), covering intended use, limits and known gaps.
- Post-market monitoring
- Monitoring a system after it has been released, to detect problems and feed back into governance.