The EU AI Act as the AIGP tests it

The EU AI Act is the one law the AIGP's Body of Knowledge names first, and competency II.C, "Understand the main elements of AI-specific laws", is written around its structure: a risk classification, a set of requirements for high-risk systems, rules on human oversight, transparency and quality management, separate rules for general-purpose AI models, an enforcement framework, and duties that differ by role. Those six indicators carry 6 to 8 of the exam's 85 scored questions. This guide takes them in order and answers each from the Act's own text, in the consolidated version of July 27, 2026 that carries the amendments of Regulation (EU) 2026/1744, with the article numbers so you can read further. The Body of Knowledge was approved in September 2025, before that amendment, so where the amendment changed something, this guide says what it was before and what it is now.

The short answer

• Six indicators, 6 to 8 questions. Competency II.C asks for the risk tiers, the high-risk requirements, oversight and transparency, general-purpose models, enforcement, and the duties by role.
• Tiers by use: prohibited practices (Article 5), high-risk systems (Article 6 and Annex III), transparency duties for certain systems (Article 50), and the rest, with voluntary codes of conduct.
• Roles decide duties: providers owe Article 16, deployers Article 26, importers and distributors verify before the market, and whoever rebrands or substantially modifies a high-risk system becomes its provider (Article 25).
• Models have their own chapter: documentation, downstream information, a copyright policy and a training-content summary, plus evaluations and incident reporting above the 10^25 FLOP threshold.
• The 2026 amendment moved the dates: the high-risk requirements apply from December 2, 2027 (Annex III) and August 2, 2028 (Annex I); fines for prohibited practices run to EUR 35 million or 7% of worldwide turnover.

What the Body of Knowledge asks

Competency II.C has six performance indicators, every one beginning "Understand":

• "the risk classification framework for AI (e.g., prohibited/high/limited/minimal risk) and what systems/uses fall into each category";
• "the key requirements around risk management, data governance, technical documentation, conformity/impact assessments and record keeping";
• "the key requirements around human oversight, transparency and notification, and quality management";
• "the distinct requirements for general-purpose AI models";
• "the enforcement framework and penalties for noncompliance";
• "the differences in requirements based on organizational context (e.g., providers, deployers, importers and distributors)".

Version 2.1 of the Body of Knowledge widened the competency to cover "other AI legislation that has now been passed besides the EU AIA", and its Domain II description names "the EU AI Act, the South Korean AI Basic Law, federal and state AI laws that apply to private sector organizations". The six indicators are written generically, so a question can ask any of them of another law, but the Act is the law whose structure they describe, and it is the one this guide covers.

First, the words the Act uses

An AI system, in Article 3(1), is "a machine-based system that is designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment, and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments". The Commission's guidelines on the definition say "That definition comprises seven main elements", the first being "a machine-based system", and the OECD's definition uses the same core words.

The roles are the Act's second vocabulary, and the sixth indicator turns on them. A provider "develops an AI system or a general-purpose AI model or that has an AI system or a general-purpose AI model developed and places it on the market or puts the AI system into service under its own name or trademark, whether for payment or free of charge". A deployer is a person or body "using an AI system under its authority except where the AI system is used in the course of a personal non-professional activity". An importer is one "located or established in the Union that places on the market an AI system that bears the name or trademark of a natural or legal person established in a third country", and a distributor is one "in the supply chain, other than the provider or the importer, that makes an AI system available on the Union market". The Act's word for all of them together, with product manufacturers and authorized representatives, is "operator".

Indicator 1: the risk tiers

The Act regulates uses, not technologies, and it sorts them into tiers.

Prohibited practices are in Article 5. The list, as consolidated, bans AI systems that deploy "subliminal techniques beyond a person's consciousness or purposefully manipulative or deceptive techniques" to distort behavior and cause significant harm; systems that exploit "the vulnerabilities of a natural person or a specific group of persons due to their age, disability or a specific social or economic situation"; social scoring that leads to detrimental treatment in unrelated contexts or that is "unjustified or disproportionate to their social behaviour or its gravity"; predicting the risk of a person committing a criminal offense "based solely on the profiling of a natural person or on assessing their personality traits and characteristics"; facial recognition databases built "through the untargeted scraping of facial images from the internet or CCTV footage"; inferring emotions "in the areas of workplace and education institutions", except for medical or safety reasons; biometric categorization that deduces "race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation"; and "real-time" remote biometric identification in publicly accessible spaces for law enforcement, with narrow exceptions. The 2026 amendment added two more: generating or manipulating sexually explicit material of an identifiable person without consent, and generating child sexual abuse material. Those two apply from December 2, 2026; the rest have applied since February 2, 2025. The Commission's guidelines on prohibited practices interpret this article, and say of themselves: "These Guidelines are non-binding."

High-risk systems are defined in Article 6 in two ways. The first is a product route: a system is high-risk where it "is intended to be used as a safety component of a product, or the AI system is itself a product, covered by the Union harmonisation legislation listed in Annex I", and that product must undergo a third-party conformity assessment. The 2026 amendment narrowed this: systems "solely used for non-safety related aspects of user assistance, performance optimisation, service efficiency, automation or convenience or quality control shall not qualify as safety components", unless their failure "would endanger health and safety". The second route is Annex III, a list of eight areas: biometrics, critical infrastructure, education and vocational training, employment and workers' management, access to essential private and public services (the Annex names creditworthiness, risk assessment and pricing in life and health insurance, and emergency call triage among them), law enforcement, "Migration, asylum and border control management", and "Administration of justice and democratic processes". An Annex III system escapes the tier where "it does not pose a significant risk of harm to the health, safety or fundamental rights of natural persons, including by not materially influencing the outcome of decision making", for example where it "is intended to perform a narrow procedural task"; a provider that concludes this must still register the system.

Limited-risk, or transparency, obligations sit in Article 50 and apply regardless of tier: a system "intended to interact directly with natural persons" must let them know they are dealing with an AI system unless that is obvious; providers of systems "generating synthetic audio, image, video or text content, shall ensure that the outputs of the AI system are marked in a machine-readable format and detectable as artificially generated or manipulated"; deployers of emotion recognition or biometric categorization systems must inform the people exposed to them; "Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake, shall disclose that the content has been artificially generated or manipulated"; and the information in all these cases is given "at the latest at the time of the first interaction or exposure".

Minimal risk has no obligations in the Act beyond the general ones; Article 95 provides for "Codes of conduct for voluntary application of specific requirements".

Indicators 2 and 3: the requirements on high-risk systems

Chapter III, Section 2 sets the requirements a high-risk system must meet, and the Body of Knowledge splits them across two indicators.

• Risk management (Article 9): "A risk management system shall be established, implemented, documented and maintained", understood as "a continuous iterative process planned and run throughout the entire lifecycle of a high-risk AI system", covering known and foreseeable risks, risks under "reasonably foreseeable misuse", and risks found through post-market monitoring.
• Data governance (Article 10): systems trained with data must use "training, validation and testing data sets" that meet the Act's quality criteria, under "data governance and management practices appropriate for the intended purpose".
• Technical documentation (Article 11): "drawn up before that system is placed on the market or put into service and shall be kept up-to date", with at least the elements of Annex IV; the amendment lets SMEs and small mid-caps use a simplified form.
• Record keeping (Article 12): the system must "technically allow for the automatic recording of events (logs) over the lifetime of the system", for traceability and post-market monitoring.
• Transparency to deployers (Article 13): the system's operation must be "sufficiently transparent to enable deployers to interpret a system's output and use it appropriately", with instructions for use.
• Human oversight (Article 14): the system must be built so that it "can be effectively overseen by natural persons during the period in which they are in use", with measures "commensurate with the risks, level of autonomy and context of use".
• Accuracy, robustness and cybersecurity (Article 15): "an appropriate level of accuracy, robustness, and cybersecurity", kept "throughout their lifecycle".

The quality management system is a provider's obligation rather than a requirement on the system: Article 16 requires providers to "have a quality management system in place which complies with Article 17". The same article lists the rest of a provider's duties: compliance with Section 2, identification on the system, documentation and logs, the conformity assessment "prior to its being placed on the market or put into service", an EU declaration of conformity, the CE marking, registration, corrective actions, demonstrating conformity on request, and accessibility.

Conformity assessment (Article 43) depends on the system. For the biometric systems of Annex III point 1, where harmonized standards exist and the provider has applied them, the provider chooses between internal control and an assessment "with the involvement of a notified body", and where the standards do not exist or were not applied, the notified-body route applies; for the other Annex III areas, providers follow "the conformity assessment procedure based on internal control as referred to in Annex VI, which does not provide for the involvement of a notified body"; and a product under Annex I follows its own sectoral procedure. Registration (Article 49) puts the provider and the system in the EU database before an Annex III system is placed on the market, critical-infrastructure systems excepted. The impact assessment the indicator names is the deployer's, below.

Indicator 6: who owes what

The Act's duties attach to roles, and the exam's scenario questions turn on them. A provider owes the list in Article 16. A deployer owes Article 26: to "take appropriate technical and organisational measures to ensure they use such systems in accordance with the instructions for use"; to "assign human oversight to natural persons who have the necessary competence, training and authority, as well as the necessary support"; where it controls the input data, to "ensure that input data is relevant and sufficiently representative"; to monitor the system, inform the provider of risks, suspend use where the system presents a risk, and report serious incidents, "first the provider, and then the importer or distributor and the relevant market surveillance authorities". Importers and distributors (Articles 23 and 24) verify before the system reaches the market.

The fundamental rights impact assessment (Article 27) is a deployer's duty, and only for some deployers: "bodies governed by public law, or are private entities providing public services", and deployers of the creditworthiness and insurance systems in Annex III. It describes the deployer's processes, the period and frequency of use, "the categories of natural persons and groups likely to be affected", the specific risks of harm, the human oversight measures, and "the measures to be taken in the case of the materialisation of those risks", and it "applies to the first use of the high-risk AI system".

Roles can change. Article 25 makes "Any distributor, importer, deployer or other third-party" a provider, with all of Article 16, when they "put their name or trademark on a high-risk AI system already placed on the market", make "a substantial modification" that leaves it high-risk, or "modify the intended purpose" of a system so that it becomes high-risk. The amendment adds that the original provider "shall no longer be considered to be a provider of that specific AI system" and must cooperate with the new one. And since the amendment, Article 4 asks both providers and deployers to "take measures to support the development of AI literacy of their staff", while saying that this "does not require providers or deployers to guarantee any specific level of AI literacy of any individual".

Indicator 4: general-purpose AI models

Chapter V is the Act's separate regime for models rather than systems. A general-purpose AI model is one "that displays significant generality and is capable of competently performing a wide range of distinct tasks regardless of the way the model is placed on the market and that can be integrated into a variety of downstream systems or applications". Its provider must, under Article 53, keep technical documentation of the model "including its training and testing process and the results of its evaluation"; give downstream providers the information they need "to have a good understanding of the capabilities and limitations of the general-purpose AI model"; "put in place a policy to comply with Union law on copyright and related rights", including the rights reservation under Article 4(3) of Directive (EU) 2019/790; and "make publicly available a sufficiently detailed summary about the content used for training", on the AI Office's template. The first two duties do not apply to models released under a free and open-source license with their parameters and architecture made public, unless the model carries systemic risk.

A model has systemic risk when it has "high impact capabilities", and it is presumed to "when the cumulative amount of computation used for its training measured in floating point operations is greater than 10^25". Its provider must, in addition, "perform model evaluation in accordance with standardised protocols and tools reflecting the state of the art, including conducting and documenting adversarial testing", assess and mitigate systemic risks at Union level, report serious incidents "to the AI Office", and "ensure an adequate level of cybersecurity protection". Codes of practice, under Article 56, are the route to showing compliance until a harmonized standard is published.

Indicator 5: enforcement and penalties

The amended Article 99 leaves it to Member States to "lay down the rules on penalties and other enforcement measures", which "shall be effective, proportionate and dissuasive", with the interests of SMEs and small mid-caps taken into account. The ceilings are the Act's own. A prohibited practice draws fines "of up to EUR 35 000 000 or, if the offender is an undertaking, up to 7 % of its total worldwide annual turnover for the preceding financial year, whichever is higher". Breaches of the obligations of providers, authorized representatives, importers, distributors and deployers, of notified bodies, and of the transparency obligations in Article 50 draw "up to EUR 15 000 000 or, if the offender is an undertaking, up to 3 %". In deciding a fine, the authorities weigh "the nature, gravity and duration of the infringement and of its consequences", the size and turnover of the operator, cooperation, "the intentional or negligent character of the infringement", and the other factors the article lists. General-purpose AI models are supervised by the AI Office, which has its own powers to request documentation, conduct evaluations and request measures, and under Article 101 "The Commission may impose on providers of general-purpose AI models fines not exceeding 3 % of their annual total worldwide turnover in the preceding financial year or EUR 15 000 000, whichever is higher"; AI systems are supervised by the Member States' market surveillance authorities.

What 2026/1744 moved, and what the exam will ask

Regulation (EU) 2026/1744, in the Official Journal of July 24, 2026, amended the Act's timetable and several of its rules. The Act "shall apply from 2 August 2026", but the high-risk requirements now start later. The amending regulation's reasoning is explicit: "it is appropriate that the date of application of Sections 1, 2 and 3 of Chapter III is set to 2 December 2027 for AI systems classified as high-risk pursuant to Article 6(2) and Annex III, and to 2 August 2028 for AI systems classified as high-risk pursuant to Article 6(1) and Annex I." The prohibitions have applied since February 2, 2025, with the two new ones from December 2, 2026; the general-purpose model chapter, the governance chapter and the penalties chapter have applied since August 2, 2025; and providers of systems that generate synthetic content placed on the market before August 2, 2026 must comply with the marking duty "by 2 December 2026". EU AI Act timeline: the dates after the Omnibus sets out every date, before and after the amendment.

The Body of Knowledge in force was approved on September 9, 2025, and the IAPP says changes to it are "communicated to candidates at least 90 days before the new content appears in the exam". Until an announcement, a question on the Act's content is safe to answer from the Act as amended, because the amendment changed dates and scope at the edges, not the structure the six indicators describe. A question that turns on a date is the one to read twice.

Studying the Act for the exam

Three habits cover most of II.C. First, classify the use, not the model: the same system is prohibited, high-risk or neither depending on what it is used for. Second, name the role before the duty: a question that describes a company usually tells you whether it developed the system, sells it under its own name, imports it or uses it, and that decides which article applies. Third, keep the model regime apart from the system regime: documentation, downstream information, copyright and the training summary belong to the model's provider, and the high-risk requirements belong to the system's.

Our AIGP course teaches II.C in six lessons that follow these indicators, each naming the articles it teaches, with practice questions in the exam's format; the first competency is free in full with a free account. The AIGP Body of Knowledge v2.1, explained sets out the competency's place in the blueprint, and AIGP exam questions: what they look like shows the shape a question on the Act takes.