Free AIGP Exam Flashcards

Memorize 50 essential terms and definitions for the IAPP Artificial Intelligence Governance Professional (AIGP). See the term, recall the definition, then flip to check yourself.

50 Flashcards
4 Topics
100% Free
TermClick to flip

Why AI needs governance beyond ordinary software

Tap to reveal definition
Card 1 of 50Foundations of AI Governance

Filter by Topic

Jump to Card

About These AIGP Flashcards

These 50 flashcards are designed to help you memorize key terms and definitions for the IAPP Artificial Intelligence Governance Professional (AIGP). Each card shows a term on the front and its definition on the back—the classic flashcard format for vocabulary memorization. Use these alongside our practice questions to build both recall and comprehension.

Topics Covered

Foundations of AI Governance11 cards
Governing AI Deployment and Use14 cards
Governing AI Development13 cards
Laws, Standards and Frameworks12 cards

Complete Flashcard Reference

Review every term in this set. Open any term to reveal its definition.

Why AI needs governance beyond ordinary software

AI outputs are probabilistic, data-dependent, and harder to predict in edge cases than deterministic software, so errors, bias, drift, opacity, and misuse can surface after release. Governance adds structured oversight, monitoring, and accountability that ordinary software controls do not cover.

Oversight for a high-volume deployment

Assign an operator who can trace harmful outputs, pause automated actions and trigger escalation. Set review thresholds based on the speed and scale at which errors could reach users.

Generative AI vs. classification model

A classifier assigns a label or score. Generative AI produces content such as text, code, audio or images. Both can make biased or unreliable outputs, and generative systems add exposure to risks such as fabricated citations and synthetic impersonation.

Agentic AI vs. retrieval-only chatbot

An agentic system can plan, call tools and chain actions. Scope its permissions and require approval before actions with significant consequences. Logging and rate limits help contain failures. A retrieval-only system still needs controls appropriate to its own data and use.

Data dependency as a governance concern

AI behavior is shaped by the quality, relevance, and representativeness of training data, so poor, biased, stale, or unrepresentative data directly distorts outputs. Governance therefore treats data provenance, quality, permissions, and fitness for purpose as first-order controls, not afterthoughts.

Fairness as a measurable control

Choose a fairness measure for the decision and test it on relevant groups when lawful and feasible. Compare selection rates or error rates, investigate the cause of disparities, and validate that the mitigation works. An observed difference can signal risk and warrants investigation.

Accountability through named owners

Accountability requires identifying who is responsible for approvals, monitoring, incidents, and remediation at each lifecycle stage. Informal or vendor-delegated ownership creates ambiguity that makes governance difficult to enforce or improve.

Safety evidence before release

Set acceptance thresholds in advance. Test under realistic failures, atypical inputs and foreseeable misuse, then compare the recorded results with the release criteria.

Human-centric AI governance

Human-centric governance keeps human rights, agency, and meaningful oversight central to design and deployment. It does not ban automation; it requires that automation remain aligned with human values and that review paths stay real, not ceremonial.

Explainability in consequential decisions

Explainability matters most when AI affects people in high-stakes ways, because the organization must understand, justify, challenge, and monitor how decisions are made. It also supports compliance analysis when outcomes are later questioned.

Deployer vs. provider in governance

Under the EU AI Act, a provider develops or commissions a system or GPAI model and supplies it under its own name. A deployer uses an AI system under its authority outside personal, nonprofessional use. One organization may have both roles.

User as a governance role

In governance terms the user is whoever interacts with the system and shapes how outputs are actually used downstream. That is why governance must train users, set acceptable-use controls, and capture feedback, because user behavior, not just model behavior, determines real-world impact.

Cloud, on-premises or edge deployment

Compare data location, latency, capacity and operational control. Cloud can simplify scaling, on-premises hosting can give the organization more infrastructure control, and edge deployment can reduce network dependence. Verify monitoring and update access for the option you choose.

Cross-functional governance group

AI risk rarely stays inside one function, so isolated legal, technical, or business review misses tradeoffs. A standing cross-functional group with legal, privacy, security, product, technical, and business representation reduces blind spots and makes ownership clearer across the lifecycle.

Procurement as a governance control point

Before adoption, procurement checks the vendor’s intended-use limits, testing, data rights, security and subcontractors. Contracts should identify who handles changes, audit evidence, incidents and exit.

Intake and review before build or deploy

At intake, record the problem, intended users, affected groups, data and accountable owner. Use that record to decide whether the design needs deeper impact assessment or a safer alternative.

Change control after a model update

Compare the update with the approved design and release evidence. Re-test affected safeguards, document validation and require approval before the changed model returns to production.

Acceptable-use policy for AI tools

Acceptable-use rules limit both the tools employees may use and the use cases they may apply them to, with explicit data-handling restrictions. Confidence or convenience is not a substitute for approval, human review, and data controls.

Live AI incident response

Contain harm and preserve logs and versions first. Then establish scope and impact, notify the responsible teams and follow applicable reporting duties before restoring the system.

IP and copyright risk in AI training

Training a model on copyrighted books or proprietary internal documents without clear permission can raise copyright and intellectual-property issues. IP review applies to third-party content and to internal material that may carry usage or confidentiality restrictions.

Discrimination law and AI screening

When an AI tool disadvantages protected groups in employment, housing, credit, or insurance, discrimination law becomes a central governance concern. A vendor source does not remove the deployer's duty to examine downstream effects.

Consumer protection and misleading AI claims

Marketing an AI system as objective or unbiased when internal testing shows disparities can create unfair or deceptive practices risk. Governance must ensure external claims match actual system behavior and evidence.

Product liability for AI-enabled products

Product liability can arise when a defect in an AI-enabled product causes covered harm. The applicable law governs what counts as a defect and compensable harm. Testing, warnings and traceability help manage that risk.

Retrieval-augmented generation or fine-tuning

Retrieval supplies approved records as context during inference. Fine-tuning changes model parameters using training examples. For frequently changing policies, retrieval can keep answers current if the system checks source access, freshness and citations.

EU AI Act risk tiers

Check Article 5 for prohibited practices, Article 6 and the annexes for high-risk classification, and Article 50 for applicable disclosures. Lower-risk systems may still be subject to cross-cutting law. Separate obligations apply to general-purpose AI models.

Human review of consequential AI outputs

Give the reviewer the underlying evidence, time, competence and authority to change the outcome. A required click without the ability to evaluate or intervene offers little oversight.

Provider duties for general-purpose AI models

Baseline EU duties include a copyright compliance policy and public training-content summary. Documentation duties have a qualified open-source exception. Models with systemic risk carry additional evaluation, mitigation, serious-incident reporting and security duties.

Transparency obligations in AI-specific law

Under Article 50, providers generally inform people when they interact directly with AI unless this is obvious in context. Providers of covered synthetic-content systems have machine-readable marking duties, and deployers of deepfakes generally disclose manipulation, subject to exceptions.

Compare open-weight and hosted models

Review license terms, access to evaluation evidence, update responsibilities and deployment control. Open weights let you operate or adapt a model, with security, monitoring and maintenance responsibilities that a managed provider may partly handle.

Lawful basis before AI training

For GDPR-covered personal data, identify an Article 6 lawful basis before processing and check Article 9 if special-category data is involved. Check purpose compatibility for reuse, limit the data to what is needed and build privacy controls in from the start.

Assess a selected AI system before roll-out

Update the impact assessment with evidence from the selected system, intended users and actual operating environment. Check affected populations, retained data, misuse paths, vendor terms and remaining risk.

Data minimization when building an AI system

Compare each training feature with the stated task. Exclude fields that are unnecessary, document the decision, and use restricted access and retention controls for personal data you need.

Localizing an AI deployment

Check the intended population, language, local legal duties and availability of review or appeal. Re-test the model and user-facing controls in that deployment context and disable uses that cannot meet local requirements.

OECD AI Principles

The OECD AI Principles are a widely cited, non-binding set of high-level expectations for trustworthy AI, including accountability, human-centered values, and transparency. They shape how organizations frame responsible AI even where they are not directly binding.

NIST AI RMF four functions

Govern sets accountability and applies across the other functions. Map frames context and impacts, Measure assesses and tracks risks, and Manage prioritizes and acts. The NIST Playbook provides suggested actions for applying the functions.

User training before production access

Teach users approved use cases, known limitations, how to verify or override outputs, which data they may enter and how to escalate problems. Restrict access to trained roles where the risks justify it.

ISO/IEC 42001

ISO/IEC 42001 specifies requirements for an auditable AI management system and can support organizational certification. ISO/IEC 22989 supplies common concepts and terminology.

Deactivating a live AI system safely

Define who can order a shutdown, the conditions that trigger it, how work moves to a fallback, and how to preserve relevant records while revoking access and isolating the system.

Use-case definition before design

Before any design work begins, governance should help define the proposed system's purpose, success criteria, and boundaries. A structured impact assessment before building surfaces legal, ethical, and technical issues while they are still cheap to fix.

Probability-severity matrix in design review

A probability-severity matrix helps teams rank AI risks by likelihood and impact so review effort can be prioritized. It turns vague risk talk into a structured design-review tool that supports go/no-go and mitigation decisions.

Data provenance vs. lineage

Data provenance records where data came from, while lineage tracks how it moves and transforms through the pipeline. Both are governance controls because they make it possible to demonstrate lawful rights, fitness for purpose, and reproducibility.

Fit for purpose in training data

Fit for purpose means the data is relevant, representative, and adequate for the intended use, not merely available. Governance must verify fitness before training rather than assuming more data always produces a better model.

Lawful rights to collect and use training data

Before training begins, the organization should document its lawful rights to collect and use the data, including licenses, consents, and contractual permissions. Undocumented rights create legal and IP exposure that may block release or trigger remediation costs.

ISO/IEC 42005 impact assessment

ISO/IEC 42005:2025 guides assessment of AI system effects on individuals, groups and society across the lifecycle. It can support an ISO/IEC 42001 management system. An organization can select an assessment method appropriate to its context.

Release readiness as a governance decision

Release readiness is a governance decision because a system can pass technical metrics yet lack monitoring, escalation paths, documentation, or user safeguards. Governance evidence, not just model metrics, is what makes release defensible.

Model card at release

A model card summarizes intended use, limitations, evaluation results, and other key details for downstream users and reviewers. It supports transparency and more disciplined downstream use than raw technical documentation alone.

Model drift and why it matters

Data drift is a change in input distribution. Concept drift is a change in the relationship between inputs and outcomes. Track both data and actual performance where labels can be obtained, then investigate before changing the model.

Red teaming in AI governance

Red teaming is structured adversarial testing that probes an AI system for failure modes standard validation tends to miss. It adds depth to readiness and monitoring evidence by viewing the system from an attacker or abuser perspective.

Threat modeling for AI

Threat modeling systematically enumerates how an AI system could be attacked or misused and which controls would fail in each path. It is most valuable when AI components introduce new entry points or incentives for abuse.

Incident logs with business and user impact

AI governance evaluates incidents by consequences to people, organizations, and systems, not only by the raw bug description. Capturing business and user impact, not just the technical defect, helps prioritize response and improve the control environment.

Frequently Asked Questions

What is the AIGP exam format?

IAPP gives candidates 165 minutes for 100 multiple-choice items, including multi-select and case-study questions, with an optional 15-minute break between the two halves. You cannot return to the first half after submitting it. Pearson VUE offers test-center and OnVUE delivery.

What score do I need to pass the AIGP?

IAPP reports AIGP results on a 100-500 scaled scoring range, and the passing score is 300. The scaled score is not the same as a raw percentage, so candidates cannot simply count correct answers to know whether they passed.

What are the four AIGP blueprint domains?

The AIGP Body of Knowledge 2.1 covers four domains: Foundations of AI Governance (16-20 scored questions), Laws, Standards and Frameworks (19-23), Governing AI Development (21-25), and Governing AI Deployment and Use (21-25). The two largest domains govern development and deployment, so candidates should weight timed practice toward lifecycle scenarios.

How much does the AIGP exam cost?

The first AIGP attempt costs USD $649 for IAPP members and $799 for nonmembers. IAPP offers discounted retakes. Certification maintenance requires 20 relevant continuing education credits per two-year term and a maintenance fee covered by membership. For AIGP nonmembers, the initial fee is included in the exam purchase and the renewal fee is $250.

What changed for AIGP prep in 2026?

IAPP’s AIGP Body of Knowledge v2.1 took effect February 2, 2026. It covers the South Korean AI Basic Law, ISO/IEC 42005 and deployment options including agentic AI. Separately, EU AI Act amendments moved the main Annex III high-risk rules to December 2, 2027 and the Annex I rules to August 2, 2028. Colorado’s SB26-189 sets new duties for covered automated decision-making technology starting January 1, 2027.

Do I need a legal or technical background before taking AIGP?

No formal prerequisite is required to sit for the AIGP exam. The blueprint is cross-functional, so candidates from privacy, legal, security, product, risk, compliance, procurement, and technical roles can all prepare successfully if they study the governance lifecycle, the major legal buckets, and current AI laws and standards.

Same family resources

Explore More IAPP Certifications

Continue into nearby exams from the same family. Each card keeps practice questions, study guides, flashcards, videos, and articles in one place.