Skip to main content
Praxikon

Subject matter and purpose of the AI Act: what Article 1 lays down and why it matters

··13 min read

Article 1 of the AI Act is short, has no obligations and is often skipped. Wrongly: its purpose counts whenever the rest of the Act gives no literal answer. Since 27 July 2026 it reads as amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744.

In brief

Paragraph 1 names four aims, paragraph 2 lists in seven points what the Regulation lays down; the Omnibus added small mid-cap enterprises to point (g). Article 1 has applied since 2 February 2025, imposes no obligation and informs interpretation. The GDPR continues to apply alongside it.

The who is in the scope of Article 2, the when in the dates of application in Article 113.

Paragraph 1: one sentence, four aims

Four aims, which sometimes pull apart:

  • Internal market. One set of rules for the Union; Member States may not impose restrictions on the development, marketing and use of AI systems, unless explicitly authorised by the Regulation (recital 1).
  • Human-centric and trustworthy AI. "Human-centric" is not defined. According to recital 6, AI should serve as a tool for people, with the ultimate aim of increasing human well-being.
  • A high level of protection of health, safety and the fundamental rights enshrined in the Charter, including democracy, the rule of law and environmental protection, against the harmful effects of AI systems (recital 5).
  • Innovation, beside protection.

Paragraph 2: seven parts, and where they are

  • (a) harmonised rules for the placing on the market, the putting into service, and the use of AI systems in the Union: with the definitions in Article 3 and AI literacy in Article 4.
  • (b) prohibitions of certain AI practices: Chapter II, Article 5.
  • (c) specific requirements for high-risk AI systems and obligations for operators of such systems: Chapter III, Article 6 to Article 49.
  • (d) harmonised transparency rules for certain AI systems: Chapter IV, Article 50.
  • (e) harmonised rules for the placing on the market of general-purpose AI models: Chapter V, Article 51 to Article 56.
  • (f) rules on market monitoring, market surveillance, governance and enforcement: Chapter VII (Article 64 to 70), the EU database in Chapter VIII (Article 71), Chapter IX (Article 72 to 94) and the penalties of Chapter XII (Article 99 to 101).
  • (g) measures to support innovation, with a particular focus on small mid-cap enterprises (SMCs) and small and medium-sized enterprises (SMEs), including start-ups: Chapter VI, Article 57 to Article 63, plus relief elsewhere (see below).

What the Digital Omnibus changed

Article 1, point (1), of Regulation (EU) 2026/1744 replaces only point (g) of paragraph 2. It originally read "with a particular focus on SMEs, including start-ups"; SMCs now stand alongside them, named first. The rest of Article 1 is unchanged.

Point (4) of the same Article adds the definitions to Article 3: SME (point (14a), by reference to Recommendation 2003/361/EC) and SMC (point (14b), by reference to point (2) of the Annex to Recommendation (EU) 2025/1099): an enterprise that is not an SME, employs fewer than 750 persons and has an annual turnover not exceeding EUR 150 million or an annual balance sheet total not exceeding EUR 129 million. According to recital 6 of Regulation (EU) 2026/1744, SMCs face challenges similar to SMEs in relation to administrative burden, and measures for small-scale operators should be extended to them where appropriate, while safeguarding the level of protection.

In the consolidated text, Articles 11, 17, 57, 70, 95, 96 and 99 mention SMCs; Articles 62 and 63 do not (see SMEs, start-ups and small mid-caps). The lower fine ceiling applies to SMCs only under Article 99(4) and (5) (paragraph 6a), not for a prohibited practice.

Article 1 is in Chapter I and has applied since 2 February 2025 (Article 113, third paragraph, point (a)); the new wording of point (g) since 27 July 2026.

Why a purpose provision matters

No one can infringe Article 1 and no fine attaches to it. It still matters.

Interpreting open norms. The Court of Justice interprets EU law by its wording, its context and the objectives of the act of which it forms part (judgment of 7 December 2023 in Case C-634/21 (SCHUFA), paragraph 41); in paragraph 51 it found its reading of Article 22 GDPR corroborated by the context and objectives of the GDPR. Open norms abound: a "significant risk of harm" (Article 6(3)), "significant changes" in design (Article 111(2)), "essential public assistance benefits and services" (Annex III, point 5). Article 1 carries weight in filling them in.

The purpose cuts both ways. According to recital 58, the Regulation should not hamper innovative approaches in the public administration, provided that those systems do not entail a high risk. In our reading, Article 1 therefore calls for a balance, not the widest or the narrowest reading. And purposive interpretation does not change clear text.

A product regulation that protects fundamental rights. Point (a) speaks the language of product legislation. Recital 3 names Article 114 TFEU, the internal market provision, as the legal basis, and Article 16 TFEU for the rules on personal data in biometrics and law enforcement; recital 9 places the Regulation in the new legislative framework. Because it also protects fundamental rights, this product law has a fundamental rights impact assessment (Article 27) and a right to explanation of individual decision-making (Article 86).

Alongside the GDPR, not above it

Under Article 2(7), the AI Act does not affect the GDPR, without prejudice to Articles 4a and 59. The Omnibus replaced the now deleted Article 10(5) there with Article 4a, the exception for processing special categories of personal data for bias detection and correction (Article 1, point (2), of Regulation (EU) 2026/1744; see bias testing).

Article 1 GDPR is a comparable purpose provision: the protection of natural persons with regard to the processing of personal data and the free movement of such data. Paragraph 2 states that the GDPR protects the fundamental rights and freedoms of natural persons, in particular the right to the protection of personal data (Article 8 of the Charter). In our reading, complying with the AI Act does not make processing of personal data lawful, and vice versa.

Example 1: the municipality

A municipality wants AI that ranks applications for debt assistance: who gets an interview first, who goes on the waiting list. Invented example.

Is this high-risk? Annex III, point 5(a), covers AI systems intended to be used by or on behalf of public authorities to evaluate the eligibility of natural persons for essential public assistance benefits and services, or to grant, reduce, revoke or reclaim them. Debt assistance is not named. According to recital 58, people applying for such services are typically dependent on them and in a vulnerable position in relation to the responsible authorities. Article 34(3) of the Charter recognises and respects the right to social assistance in order to combat social exclusion and poverty. In our reading, debt assistance is therefore an essential public service. Whether this system falls under point 5(a) depends on what the ranking does: point 5(a) covers evaluating eligibility and granting or reducing, not setting an order (Annex III names prioritisation only in point 5(d)). If the ranking in practice decides who is helped and who is not, or only much later, the system is in our reading in principle high-risk. Article 6(3) then does not help: a ranking of residents based on their personal circumstances is profiling, and an Annex III system that performs profiling is always high-risk.

What then follows. If the system is put into service after 2 December 2027, the deployer obligations apply (Article 26), including human oversight by persons with the necessary competence (paragraph 2), the fundamental rights impact assessment before first use (Article 27(1)) and registration (Article 26(8) and Article 49(3)). If the same type and model was already placed on the market or put into service earlier (recital 39 of Regulation (EU) 2026/1744), Article 111(2) applies those obligations only after significant changes in its design, and for a system intended to be used by public authorities by 2 August 2030. If it builds the system and puts it into service under its own name, it is also the provider (Article 3, points (3) and (4)). In our reading, "human-centric" here means that staff can override the ranking.

What already applies. Article 4 and the prohibition of social scoring (Article 5(1)(c)); in our reading that prohibition does not catch a ranking within debt assistance itself, as long as it does not lead to unjustified or disproportionate treatment. Then the GDPR: a legal basis, usually a task carried out in the public interest (Article 6(1)(e) GDPR), and a data protection impact assessment where a high risk is likely (Article 35 GDPR). Article 22 GDPR requires a decision, based solely on automated processing, with legal or similarly significant effects (SCHUFA, paragraph 43). If the ranking largely determines who is helped without any review on the merits by staff, it may in our reading be such a decision; in SCHUFA the Court held so for a credit bureau's score on which a bank draws strongly to establish a contractual relationship. Genuine review by a staff member takes the decision outside Article 22 GDPR. See AI welfare fraud risk scoring.

Example 2: the bank and the insurer

A bank buys a consumer credit model from a supplier with 400 employees (invented).

  • The text gives the answer. Credit scoring of natural persons is in principle high-risk (Annex III, point 5(b)). From 2 December 2027 the bank carries out a fundamental rights impact assessment (Article 27(1)); for a model already in use before that date, Article 111(2) applies.
  • Point (g) in practice. If the supplier is an SMC, it may provide the Annex IV elements in a simplified manner using the form the Commission is to establish for that purpose (Article 11(1)), and its quality management is proportionate to its size (Article 17(2)).
  • The limit of purposive interpretation. An insurer uses AI to assess claims under a health insurance policy. Annex III, point 5(c), covers risk assessment and pricing, not claims assessment. In our reading, Article 1 does not make this high-risk; the Commission can do so through Article 7, or the Union legislature by amending the Regulation. Article 4 does apply, Article 50 only if the system, for example, interacts directly with policyholders (paragraph 1), and the GDPR applies in full; Article 22 GDPR comes into play if a claim is rejected solely by automated means.

What to do now

  1. Record how Article 1 and the recitals support your reading where the text is silent.
  2. Note in your AI register whether you or your supplier is an SME or an SMC.
  3. Run two tracks: a data protection impact assessment where Article 35 GDPR requires one, and a fundamental rights impact assessment if you are a public body or provide public services, or assess credit or life and health insurance (Article 27(1)), from the date that applies to you; for existing systems with Article 111(2).

The case check shows which duties apply to your use case.

Frequently asked questions about Article 1 of the AI Act

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

Related obligations

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

When does the AI Act apply? Article 113 and the dates of application after the Digital Omnibus

Article 113 of the AI Act sets out when each part of the Regulation applies. The AI Act has been in force since 1 August 2024, but it applies in stages, from 2 February 2025 to 2 August 2028. All dates as they stand after Regulation (EU) 2026/1744, what the Digital Omnibus changed, how Article 111 deals with existing systems, and what a municipality and a bank must already do.

EU AI ActArticle 113Dates of application

Deployer obligations for high-risk AI: what Article 26 of the AI Act requires

An organisation that uses a high-risk AI system has twelve paragraphs of obligations of its own (Article 26 AI Act): use in accordance with the instructions, human oversight by competent people, sound input data, monitoring and notification, keeping logs, informing workers and affected persons, registration by public authorities, feeding the DPIA and cooperating with the authorities. What each paragraph requires, what the Digital Omnibus did and did not change, the rules for banks and insurers, the difference with the provider, and how the FRIA and the right to explanation connect, with a municipality, a bank and an employer as examples.

EU AI ActArticle 26Deployer

The quality management system for high-risk AI: what Article 17 of the AI Act asks of providers

A provider of a high-risk AI system must have a quality management system that ensures compliance with the AI Act, documented in written policies, procedures and instructions (Article 17). That obligation rests on the provider, not on the organisation that only uses a system it has bought. What the thirteen aspects of paragraph 1 require, what the Digital Omnibus changed for SMEs and small mid-caps, how sectoral quality systems and bank governance count, what EN 18286:2026 does and does not prove, and how it connects with Articles 9, 10, 11, 12, 72 and 73, with a municipality that builds its own system and a bank as examples.

EU AI ActArticle 17Quality management