Skip to main content
Praxikon

Does the AI Act apply to AI in cars, aircraft, trains and ships? Articles 102 to 110 and Section B of Annex I

··32 min read

A municipality has a self-driving shuttle run across a business park. A province requires buses with driver assistance systems in a new bus concession. A leasing company buys cars that brake by themselves when a cyclist crosses. Does all that AI fall under the AI Act? The answer is spread over four places in the Regulation: Article 2(2), Article 6(1), Section B of Annex I and Articles 102 to 110. Together they say: yes, but not in the way you might expect. For AI that is a safety component of such a product, only a few articles of the AI Act itself apply. The requirements for high-risk AI systems come in through the product legislation of that sector. The examples in this piece are fictitious.

In short

Section B of Annex I lists nine pieces of Union legislation: on civil aviation security, two- or three-wheel vehicles and quadricycles, agricultural and forestry vehicles, marine equipment, the rail system, motor vehicles (two acts), civil aviation and, since the Digital Omnibus on AI (Regulation (EU) 2026/1744), machinery.

Where an AI system is classified as high-risk under Article 6(1) and is related to a product covered by one of those acts, only Article 6(1), Article 60a and Articles 102 to 112 of the AI Act apply. Articles 57, 58 and 59 on AI regulatory sandboxes apply only in so far as the requirements of the AI Act have been integrated in that sectoral legislation (Article 2(2), as replaced by the Omnibus).

Articles 102 to 109 amend eight of those acts. Each time the text is the same: when the Commission adopts delegated or implementing acts under that act concerning AI systems which are safety components, the requirements set out in Chapter III, Section 2, of the AI Act shall be taken into account. Those are the requirements of Articles 8 to 15. For machinery, the Omnibus deals with it in the Machinery Regulation itself, with a fixed date: the delegated acts shall apply by 2 August 2028.

Article 110 is something else. It adds the AI Act to the list in the Union directive on representative actions, so that qualified entities can bring actions on behalf of consumers over infringements of the AI Act.

What still applies: the AI Act for all other AI in the same organisation and for AI in the vehicle that is not a safety component, and always the GDPR where personal data are processed.

This piece is about the products of Section B. For AI in products under Section A, such as medical devices and toys, the AI Act does apply in full, alongside the product legislation. See medical AI under the AI Act and the MDR and conformity assessment: internal control or a notified body. Testing on public roads or in the air is covered in testing in real-world conditions; for the products of Section B only Article 60a of those rules applies directly.

Which acts are in Section B?

Annex I consists of two lists. Section A is the "List of Union harmonisation legislation based on the New Legislative Framework". Section B is the "List of other Union harmonisation legislation". After the Omnibus, Section B contains the following acts, with the article of the AI Act that amends them:

PointActSectorAmended by
13Regulation (EC) No 300/2008Civil aviation securityArticle 102
14Regulation (EU) No 168/2013Two- or three-wheel vehicles and quadricyclesArticle 104
15Regulation (EU) No 167/2013Agricultural and forestry vehiclesArticle 103
16Directive 2014/90/EUMarine equipmentArticle 105
17Directive (EU) 2016/797Interoperability of the rail systemArticle 106
18Regulation (EU) 2018/858Approval and market surveillance of motor vehicles and their trailersArticle 107
19Regulation (EU) 2019/2144Type-approval of motor vehicles as regards their general safetyArticle 109
20Regulation (EU) 2018/1139Civil aviation and EASAArticle 108, and Article 2 of Regulation (EU) 2026/1744
21Regulation (EU) 2023/1230MachineryArticle 3 of Regulation (EU) 2026/1744

Two points deserve attention.

Aviation is limited. Point 20 does not list Regulation (EU) 2018/1139 as a whole, but "in so far as the design, production and placing on the market of aircrafts referred to in Article 2(1), points (a) and (b) thereof, where it concerns unmanned aircraft and their engines, propellers, parts and equipment to control them remotely, are concerned". In our reading this means drones. The sentence is not entirely unambiguous. If our reading is right, AI in manned aircraft is not high-risk via Annex I and does not fall under the limitation of Article 2(2) either. On the other hand, the amendments to the aviation regulation made by Article 108 and by the Omnibus clearly go beyond drones, covering airworthiness, air operations and air traffic controllers. Anyone building AI for manned aviation should therefore put this question to the aviation authority.

Machinery has moved. Until 27 July 2026 the machinery legislation was in Section A, as point 1 (the old Machinery Directive 2006/42/EC). The Omnibus deleted that point and added the Machinery Regulation (EU) 2023/1230 as point 21 of Section B. According to recital 42 of the Omnibus, this is a move to a sectoral approach, because applying both regulations side by side might lead to overlaps. AI that is a safety component of a robotic arm, an autonomous forklift or a mowing machine has followed the Section B route since then, where a third-party conformity assessment applies to that machine (Article 6(1), point (b)). The Machinery Regulation itself applies from 14 January 2027.

When is AI in such a product high-risk?

That is governed by Article 6(1), the only classification provision that remains for Section B. An AI system is considered to be high-risk where both of the following conditions are fulfilled:

  1. it is intended to be used as a safety component of a product covered by an act listed in Annex I, or it is itself such a product (point (a));
  2. that product is required to undergo a third-party conformity assessment under that act with a view to its placing on the market or putting into service (point (b)).

What a safety component is. The Omnibus replaced the definition. A safety component is now a component of a product or of an AI system which fulfils a safety function for that product or AI system, or the failure or malfunctioning of which endangers the health and safety of persons or property. A component fulfils a safety function where its intended purpose is to prevent or mitigate risks to health and safety of persons or property (Article 3, point (14)).

What is not a safety component. The Omnibus also added three paragraphs to Article 6. AI systems that are solely used for non-safety related aspects of user assistance, performance optimisation, service efficiency, automation or convenience or quality control do not qualify as safety components (paragraph 1a). But AI systems the failure or malfunctioning of which would endanger health and safety do qualify (paragraph 1b). And a product that requires third-party conformity assessment solely due to risks other than risks to health and safety, such as the distribution of radio spectrum or electromagnetic interference, does not fulfil condition (b) (paragraph 1c).

In our reading, in a car this means: an emergency braking system that detects a cyclist and brakes fulfils a safety function. A voice assistant that picks the music or suggests a route based on the driver's calendar is meant for convenience and is not a safety component, as long as its failure endangers no one. That voice assistant therefore does not follow the Section B route, and the AI Act simply applies to it.

What applies from the AI Act: Article 2(2)

The core provision is short. For AI systems classified as high-risk under Article 6(1) related to products covered by Section B, "only Article 6(1), Article 60a and Articles 102 to 112 shall apply. Articles 57, 58 and 59 shall apply only in so far as the requirements for high-risk AI systems under this Regulation have been integrated in that Union harmonisation legislation."

ProvisionSubjectApplies for Section B
Article 6(1)Classification as high-riskYes
Article 60aTesting in real-world conditions outside AI regulatory sandboxesYes, if the Member State adopts a framework
Articles 102 to 110Amendments to the sectoral acts and to Directive (EU) 2020/1828Yes
Articles 111 and 112Systems already placed on the market; evaluation and reviewYes
Articles 57, 58 and 59AI regulatory sandboxesOnly in so far as the requirements have been integrated in the sectoral act
Articles 8 to 15, 16 to 27, 43, 49, 72 and 73, Chapter IXRequirements, obligations of providers and deployers, conformity assessment, registration, monitoring, incident reporting, market surveillanceNot directly

What the Omnibus changed. In the original 2024 text only Article 6(1), Articles 102 to 109 and Article 112 applied, and only Article 57 in so far as the requirements had been integrated in the sectoral act. Since 27 July 2026, Article 60a, Articles 110 and 111 and Articles 58 and 59 have been added.

What "only" means. The deployer of such a system therefore has no obligations under Article 26, there is no fundamental rights impact assessment under Article 27 and no registration in the EU database under Article 49. The provider does not follow the conformity assessment of Article 43, but the approval or assessment procedure of the sectoral act. According to recital 27 of the Omnibus, these systems are subject to the requirements and procedures of the relevant sectoral legislation and are "for most purposes, not directly subject to" the AI Act. In our reading, at least once Article 6(1) applies to Annex I on 2 August 2028, Article 4 on AI literacy and the transparency obligations of Article 50 therefore do not apply directly to such a system either. For the period before that date this is not certain: Article 4 has applied since 2 February 2025 and Article 50 since 2 August 2026, while the limitation of Article 2(2) is tied to a classification that only works from 2 August 2028. Whether the limitation also sets aside the prohibitions of Article 5 is, in our reading, not certain either. For a system that monitors the driver, assume that the prohibition of emotion recognition in the workplace continues to apply. Those provisions may well apply to the other AI of the same organisation.

Supervision. Because Chapter IX does not apply, no market surveillance authority under the AI Act supervises these systems. That is done by the authorities under the sectoral act: the approval authorities and market surveillance authorities that Member States have to establish or appoint under Article 6 of Regulation (EU) 2018/858, and in aviation the Agency and the authorities under Regulation (EU) 2018/1139. That differs from Section A, where the authority under the product legislation is as a rule also the market surveillance authority for the AI Act (Article 74(3)).

How the requirements come in: Articles 102 to 109

Articles 102 to 109 add a short provision to each sectoral act. Recital 49 of the AI Act gives the purpose: the Commission should take into account, "on the basis of the technical and regulatory specificities of each sector, and without interfering with existing governance, conformity assessment and enforcement mechanisms and authorities established therein", the mandatory requirements for high-risk AI systems when adopting any relevant delegated or implementing acts on the basis of those acts.

AI Act articleWhat it adds
102To Article 4(3) of Regulation (EC) No 300/2008: when adopting detailed measures related to technical specifications and procedures for approval and use of security equipment concerning AI systems
103To Article 17(5) of Regulation (EU) No 167/2013: when adopting delegated acts concerning AI systems which are safety components
104To Article 22(5) of Regulation (EU) No 168/2013: the same
105A new Article 8(5) of Directive 2014/90/EU: for the activities of the Commission and when adopting technical specifications and testing standards
106A new Article 5(12) of Directive (EU) 2016/797: when adopting delegated acts and implementing acts
107A new Article 5(4) of Regulation (EU) 2018/858: when adopting delegated acts amending the list of regulatory acts with technical requirements in Annex II to that regulation
108New paragraphs in Articles 17, 19, 43, 47 and 58 and a new subparagraph in Article 57 of Regulation (EU) 2018/1139: airworthiness, air traffic management and air navigation services, and unmanned aircraft
109A new Article 11(3) of Regulation (EU) 2019/2144: when adopting implementing acts on automated and fully automated vehicles

Four things stand out.

"Taken into account" is not a copy. The text does not say that the requirements of Articles 8 to 15 are inserted word for word into the sectoral act. The Commission takes them into account. Recital 27 of the Omnibus says: "Those sectoral acts will, in due course, incorporate requirements corresponding to the requirements set out in Articles 8 to 15 of that Regulation." Until that has happened, in our reading those requirements do not apply as such to the manufacturer; what applies is what the sectoral act and the acts adopted under it require.

No deadline, except for machinery. Articles 102 to 109 set no date by which the Commission has to adopt those acts. This piece describes what the AI Act provides; which acts the Commission has already adopted or is preparing under these sectoral acts, and to what extent they already reflect the requirements, we have not examined here. Check the sectoral act itself.

Automated vehicles. Article 109 touches the provision on automated and fully automated vehicles. Under Article 11 of Regulation (EU) 2019/2144 the Commission adopts implementing acts on, among other things, systems that replace the driver's control of the vehicle, systems that provide the vehicle with real-time information on its state and surroundings, driver availability monitoring systems and event data recorders. That is where a lot of AI sits.

A difference between language versions. In Article 105 the Dutch text refers to "hoofdstuk III, hoofdstuk 2" (Chapter III, Chapter 2). The English text says "Chapter III, Section 2", like all the other articles. In our reading Section 2 of Chapter III is meant. And Article 102 is worded differently from the rest: Regulation (EC) No 300/2008 is about aviation security, and the provision refers to security equipment concerning AI systems, not to safety components.

What the Omnibus added for aviation and machinery

Aviation. According to recital 44 of the Omnibus, a technical correction was necessary so that the requirements "are fully covered". Article 2 of Regulation (EU) 2026/1744 therefore added the same sentence to seven further articles of Regulation (EU) 2018/1139, on the training, testing, checking and medical assessment of pilots, cabin crew and air traffic controllers, on air operations and on aerodromes.

Machinery. Article 3 of Regulation (EU) 2026/1744 goes further than Articles 102 to 109. It requires the Commission to add, by delegated act, health and safety requirements to Annex III to the Machinery Regulation for AI systems classified as high-risk under Article 6(1) because they are a safety component in a product covered by that regulation, or are themselves such a product. Those requirements reflect the requirements of Chapter III, Section 2, and Articles 17, 19, 72 and 73 of the AI Act: quality management, logs, post-market monitoring and the reporting of serious incidents. The delegated acts shall apply by 2 August 2028 (Article 8 of Regulation (EU) 2023/1230, as supplemented). Until harmonised standards or common specifications for AI are referenced or adopted under the Machinery Regulation itself, high-risk AI systems under that regulation that comply with the harmonised standards or common specifications of the AI Act are presumed to conform with those requirements (Article 20(10) of Regulation (EU) 2023/1230).

Evaluation and testing

Article 112(12). Any amendment to the AI Act following an evaluation, and any relevant delegated or implementing acts, which concern the sectoral legislation of Section B shall take into account the regulatory specificities of each sector, the existing governance, conformity assessment and enforcement mechanisms, and the authorities established therein.

Article 60a. Member States may allow providers to test these systems in real-world conditions outside AI regulatory sandboxes, in order to assess and verify their conformity with Articles 8 to 15. A Member State that chooses to do so adopts a framework and notifies it to the Commission before implementing it. The testing must comply with the applicable provisions of the sectoral act. Whether the Netherlands will adopt such a framework is not yet settled; the draft Dutch AI Act Implementation Act predates the Omnibus and does not deal with Article 60a.

Article 110: representative actions on behalf of consumers

Article 110 belongs to this group, but has nothing to do with product legislation. It adds the AI Act as point 68 to Annex I of Directive (EU) 2020/1828 on representative actions for the protection of the collective interests of consumers. That directive applies to representative actions brought against infringements by traders of the provisions listed in its Annex I that harm or may harm the collective interests of consumers. A trader there is any natural or legal person, whether privately or publicly owned, acting for purposes relating to its trade, business, craft or profession. In our reading, a qualified entity can therefore bring an action on behalf of consumers against a provider or deployer that is a trader and, for example, engages in a prohibited practice, where that harms or may harm the collective interests of consumers. Consumers do not bring such an action themselves. How it proceeds is set out in the national law transposing the directive. For the systems of Section B this means little, because the substantive requirements there come from the sectoral act.

From when?

  • Articles 102 to 110 apply from 27 July 2026 (Article 113, third paragraph, point (d), inserted by the Omnibus). The original text had no separate date for them, and the general date of 2 August 2026 applied.
  • Article 2(2) is in Chapter I and has applied since 2 February 2025, in its current wording since 27 July 2026.
  • Article 6(1) is in Section 1 of Chapter III. For AI systems classified as high-risk under Article 6(1) and Annex I, that Section applies from 2 August 2028 (Article 113, third paragraph, point (c)(ii)). The text does not distinguish between Section A and Section B of Annex I here.
  • Article 60a is in Chapter VI. That Chapter is not among the exceptions in Article 113 and falls under the general date of 2 August 2026.
  • Machinery: the delegated acts under the Machinery Regulation shall apply by 2 August 2028.

What still applies

Other AI in the same organisation. Article 2(2) only concerns the AI system that is a safety component. A transport operator, municipality or leasing company that uses such a vehicle remains subject to the AI Act for all its other AI: a chatbot for passengers, a system that assigns shifts to drivers, a model that assesses damage claims. A few boundaries from the text:

  • Traffic management is Annex III. AI intended to be used as a safety component in the management and operation of road traffic is listed in point 2 of Annex III and is in principle a high-risk AI system, with the deployer obligations of Article 26. For that point there is no fundamental rights impact assessment (Article 27(1)) and no registration by the deployer in the EU database (Article 49(3)). That is something other than AI in the vehicle itself.
  • Work and staff. AI used to allocate tasks based on individual behaviour or personal traits or characteristics, or to monitor and evaluate the performance and behaviour of workers, is listed in point 4(b) of Annex III.
  • Emotion recognition in the workplace is prohibited, except where the use of the AI system is intended to be put in place or into the market for medical or safety reasons (Article 5(1), point (f)). According to recital 18, the notion of emotion recognition does not include physical states such as pain or fatigue, including systems used in detecting the state of fatigue of professional pilots or drivers for the purpose of preventing accidents.
  • Interaction with people. The provider of a chatbot ensures that people are informed that they are interacting with an AI system, unless this is obvious (Article 50(1)), and an emotion recognition system that does fall under the AI Act brings an information obligation (Article 50(3)).

The GDPR. Article 2(2) says nothing about personal data. Cameras in and around the vehicle, systems that monitor the driver's attention and event data recorders often process personal data, including those of passers-by. Recital 14 of Regulation (EU) 2019/2144 states that the processing of information about the driver in event data recorders and in driver drowsiness and attention warning or advanced driver distraction warning systems should always be carried out in accordance with Union data protection law, in particular the GDPR. According to the same recital, those warning systems should not continuously record or retain data other than what is necessary within the closed-loop system. The binding rule is in Article 6(3) of Regulation (EU) 2019/2144: those data shall not be accessible or made available to third parties at any time and shall be immediately deleted after processing. In our reading an employer therefore cannot use the data from that warning system for its own purposes. Anyone who collects data with a separate system of its own, for example an employer analysing the driving behaviour of drivers, needs its own legal basis (Article 6 GDPR). Where that processing is likely to result in a high risk, the controller carries out a data protection impact assessment beforehand (Article 35 GDPR(1)). Whether that is the case depends on the processing; the AI Act does not decide it.

Main example: the municipality with a shuttle and the province with a bus concession

The shuttle. In 2029 a municipality wants a transport operator to run a self-driving shuttle between a station and a business park. What is its role?

  1. Does the AI in the shuttle fall under Section B? In our reading it does. The system that takes over driving fulfils a safety function and belongs to a motor vehicle that requires approval by an approval authority under Regulation (EU) 2018/858 and Regulation (EU) 2019/2144. Then Article 6(1) applies and, through Article 2(2) of the AI Act, hardly anything else.
  2. Is the municipality a provider or a deployer within the meaning of the AI Act? For the driving software that question matters little in our reading: the obligations of providers and deployers in Chapter III do not apply to this system. The manufacturer handles the approval of the vehicle under the sectoral act.
  3. Does the shuttle run before it has been approved? Then it is testing. In our reading the AI Act does not itself regulate such testing here, except that a Member State may, under Article 60a, permit providers or prospective providers to test in real-world conditions. Whether the shuttle may run follows from the sectoral act and national traffic law. This piece does not cover those national rules.
  4. What the municipality does have to arrange. The shuttle's cameras and sensors record people in the street. Who the controller is follows from who determines the purposes and means of the processing (Article 4(7) GDPR). The municipality records that with the operator and the manufacturer, and in the case of joint controllers in an arrangement between them (Article 26 GDPR). The GDPR applies to whoever is the controller, with, in our reading, a good chance that a DPIA is required. An app or chatbot through which passengers book the shuttle falls under the AI Act as usual, including Article 50(1).
  5. And the traffic lights? If the municipality uses AI to control traffic lights near the stop, that can be a safety component in the management of road traffic (Annex III, point 2). That is in principle a high-risk AI system under the AI Act, with obligations for the municipality as deployer from 2 December 2027.

The bus concession. A province grants the concession for regional bus services and requires the buses to have driver assistance systems. In our reading, the province as the granting authority is neither the provider nor the deployer of those systems: the manufacturer places them on the market and the operator uses them. And because they are safety components of vehicles, the obligations of Article 26 do not apply. What the province can do is ask in the concession for the approval of the vehicles and for arrangements on the data the systems collect. If the province uses AI itself, for example to filter applications for its own vacancies, that is in principle a high-risk AI system under point 4(a) of Annex III. Then, as a public authority, it also has to carry out the fundamental rights impact assessment (Article 27(1)) and register before use (Article 49(3)) from 2 December 2027, and it may not use a system that has not been registered in the EU database (Article 26(8)). For a system placed on the market or put into service before that date, the rules apply only after a significant change in its design; if it is intended to be used by public authorities, it must comply by 2 August 2030 at the latest (Article 111(2)).

Example 2: the leasing company, the bank and the insurer

The lease car. A leasing company buys cars with automated emergency braking and lane keeping assistance. In our reading it has no obligations as deployer of a high-risk AI system for those systems: they are safety components of vehicles under Section B. If the leasing company also offers an app that scores driving behaviour and gives the driver a discount, that is a separate AI application, independent of the vehicle. It falls under the AI Act and, because it concerns the driving behaviour of an identifiable driver, under the GDPR.

The car loan. A bank that uses AI to evaluate the creditworthiness of a consumer who wants to finance a car is in principle using a high-risk AI system (Annex III, point 5(b)). The bank is then the deployer and also performs a fundamental rights impact assessment (Article 27(1)). That it concerns a car changes nothing.

The car insurance. Point 5(c) of Annex III covers risk assessment and pricing in the case of life and health insurance. An insurer that sets the premium of a car insurance with telematics data does not, in our reading, fall under that point. It is then not high-risk, unless another point of Annex III applies. The GDPR does apply, and for a decision based solely on automated processing which produces legal effects concerning a person or similarly significantly affects them, Article 22 GDPR as well.

Example 3: the university of applied sciences with a drone

A university of applied sciences develops, with students, a drone that avoids obstacles using AI, and tests it on its own site. As long as the drone is not placed on the market or put into service, research, testing and development fall outside the AI Act, except testing in real-world conditions (Article 2(8)). If the drone becomes a product, in our reading it falls under point 20 of Section B. The route of the aviation legislation applies where a third-party conformity assessment is required for the drone (Article 6(1), point (b)); without that assessment the system is not high-risk via Annex I and the limitation of Article 2(2) does not apply. If the drone films people, the GDPR applies. Separately, the university's AI for admission, assessment and proctoring falls in principle under point 3 of Annex III.

Example 4: the transport operator as employer

A bus operator has buses with a driver drowsiness and attention warning, one of the systems that Article 6 of Regulation (EU) 2019/2144 requires for motor vehicles. That system warns the driver and is part of the vehicle. In our reading it follows the Section B route. If the operator also wants its own camera system that assesses the behaviour of drivers and takes it into account in their appraisal, that is something else:

  1. High-risk. Monitoring and evaluating the performance and behaviour of workers is listed in point 4(b) of Annex III; the system is then in principle a high-risk AI system. The operator is then the deployer, with the obligations of Article 26 from 2 December 2027, including informing workers' representatives and the affected workers before putting the system into service (Article 26(7)).
  2. Emotion recognition. If the system infers emotions, such as anger, on the basis of biometric data (Article 3(39)), that is prohibited in the workplace unless the use is intended for medical or safety reasons (Article 5(1), point (f)). Detecting fatigue to prevent accidents is not emotion recognition according to recital 18.
  3. GDPR. The operator needs a legal basis, and for a systematic assessment of employees a DPIA is in our reading the obvious step (Article 35 GDPR).

What to do now

  1. List the AI in and around your vehicles, aircraft, ships, trains and machinery, and record for each system whether it is a safety component or not. Use the definition in Article 3, point (14), and Article 6(1a) and (1b).
  2. Safety component under Section B? Ask the manufacturer for the approval under the sectoral act. Do not expect an EU declaration of conformity or a registration under the AI Act.
  3. Not a safety component? Then the AI Act applies as usual. Check whether the system is listed in Annex III and whether Article 5 or Article 50 plays a role.
  4. Look at your own AI around the vehicle: traffic management, staff planning, assessment of drivers, customer contact.
  5. Sort out the GDPR: who is the controller for the camera and sensor data, on what legal basis, and is a DPIA required?
  6. Working with machinery? Take into account the new requirements in Annex III to the Machinery Regulation, which shall apply by 2 August 2028.

Who will supervise which AI systems in the Netherlands is covered in who will supervise the AI Act. The draft explanatory memorandum to the draft Dutch AI Act Implementation Act states that the authorities supervising the legislation of Section B are not designated as market surveillance authority under the AI Act (internet consultation, in Dutch). The draft is not yet law.

Want to know whether an application falls under Annex III and which obligations under the AI Act and the GDPR go with it? With the case check you describe the application in a few questions.

Frequently asked questions about AI in cars, aircraft, trains and ships

Sources

EUR-Lex: Directive (EU) 2020/1828 on representative actions, Articles 1, 2 and 3 (25 November 2020, accessed 19 November 2026)
EUR-Lex: Regulation (EU) 2016/679 (GDPR), Articles 6, 22 and 35 (27 April 2016, accessed 19 November 2026)
Ministry of Economic Affairs and Climate Policy: Internet consultation on the Dutch AI Act Implementation Act: draft bill and draft explanatory memorandum (passages on Section B of Annex I), in Dutch (20 April to 1 June 2026, accessed 19 November 2026)

Newsletter

Every Tuesday, the AI Act week ahead in 5 minutes

A practical briefing on deadlines, new guidance and enforcement, so you know what matters this week. No spam and you can unsubscribe in one click.

Practical and short · No spam · One-click unsubscribe

Written by

Zahed Ashkara, freelance AI and privacy consultant and legal adviser

Zahed Ashkara

Freelance AI and privacy consultant and legal adviser · AI governance & safety

Zahed Ashkara is a freelance AI and privacy consultant and legal adviser for the GDPR and the EU AI Act. Organisations hire him as AI governance consultant, privacy counsel, external DPO or AI governance lead, have him carry out a DPIA, FRIA or AI Act classification, or book an in-company training. He focuses on AI governance and safety, including AI agents. He works exclusively as a freelancer, through Embed AI (trade name Praxikon). He is a Certified AI Compliance Officer (CAICO) and has been a member of the NEN standards committee Artificial Intelligence & Big Data since July 2026.

Follow on LinkedInView expert profile

Use these sources in your AI assistant too: Praxikon in Claude and other assistants

Real-World Stories

Discover how organizations apply this in practice

Related articles

Certificates and the emergency derogation: what Articles 44, 45 and 46 of the AI Act regulate

A certificate from a notified body is valid for four or five years at most and can be suspended or withdrawn. And very occasionally a high-risk AI system may be used while the conformity assessment is still under way. What Articles 44, 45 and 46 of the AI Act regulate, who will decide on such an authorisation in the Netherlands, why the GDPR simply continues to apply, and what a safety region, a municipality and a bank need to do.

EU AI ActArticle 44Article 45

Codes of conduct under Article 95 of the AI Act: voluntary rules for AI that is not high-risk

Article 95 of the AI Act lets organisations draw up codes of conduct on a voluntary basis: to apply the requirements for high-risk systems to AI that is not high-risk, and to set their own objectives on ethics, the environment, AI literacy, inclusive design and vulnerable groups. What such a code does and does not do, who may draw it up, how it differs from the codes of practice under Articles 50 and 56 and from codes of conduct under the GDPR. With a partnership of municipalities as the main example.

EU AI ActArticle 95Codes of conduct

Reporting an AI Act infringement as a whistleblower: Article 87 and the Dutch Whistleblower Protection Act

Article 87 of the AI Act makes the EU Whistleblower Directive apply to reports of infringements of the AI Act. What that means in the Netherlands under the Whistleblower Protection Act: who is protected, when an employer needs an internal reporting procedure, where you can report externally, what is not yet settled and where the GDPR comes in. With a municipal civil servant and a bank employee as examples.

EU AI ActArticle 87Whistleblowers