High-risk AI systems an SME may be using without knowing it (Annex III)
28 October 2026 · Sources checked on 28 October 2026.
The EU AI Act (Regulation (EU) 2024/1689) calls two groups of systems "high-risk" (Article 6): those that are safety components of products covered by the Union legislation in Annex I, or are those products themselves (toys, machinery, medical devices, personal protective equipment, among others), and those that fall within one of the eight areas of Annex III. The first are made by industry; the second are often management software that an SME contracts without thinking about the Regulation: a CV filter, a tool that allocates tasks or evaluates performance, a system that scores customers' creditworthiness.
The high-risk rules (Chapter III, Sections 1 to 3) do not apply yet. After Regulation (EU) 2026/1744, Article 113, third paragraph, point (c), sets two dates: 2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems. This article summarises the official text as it applies today; it does not interpret and it does not constitute legal advice.
The eight Annex III areas
| Point | Area | Which systems it includes (summary of the official text) |
|---|---|---|
| 1 | Biometrics, in so far as their use is permitted | Remote biometric identification (not verification whose sole purpose is to confirm that a person is who they claim to be); biometric categorisation by sensitive or protected attributes; emotion recognition |
| 2 | Critical infrastructure | Safety components in the management and operation of critical digital infrastructure, road traffic, or the supply of water, gas, heating or electricity |
| 3 | Education and vocational training | Determining access or admission to educational and vocational training institutions; evaluating learning outcomes; assessing the level of education a person will receive; monitoring prohibited behaviour during tests |
| 4 | Employment, workers' management and access to self-employment | Recruitment or selection (placing targeted job advertisements, analysing and filtering applications, evaluating candidates); decisions on terms of work-related relationships, promotion or termination, allocating tasks based on individual behaviour or personal traits, and monitoring and evaluating performance and behaviour |
| 5 | Access to and enjoyment of essential private services and essential public services and benefits | Evaluating eligibility for public assistance benefits (by or on behalf of public authorities); evaluating the creditworthiness or credit score of natural persons (except for detecting financial fraud); risk assessment and pricing in life and health insurance; evaluating and classifying emergency calls, dispatching emergency services and patient triage |
| 6 | Law enforcement | Systems used by or on behalf of law enforcement authorities for the purposes the point lists |
| 7 | Migration, asylum and border control management | Systems used by or on behalf of the competent public authorities for the purposes the point lists |
| 8 | Administration of justice and democratic processes | Assisting a judicial authority in researching and interpreting facts and the law (or in alternative dispute resolution); influencing the outcome of an election or referendum or voting behaviour |
The table summarises; the text that counts is the official one, linked below.
The ones closest to an SME's day-to-day work
Points 6 and 7 concern public authorities, and points 2 and 8 very specific activities. The ones that can turn up in any company are these:
- **Point 4, employment.** Recruitment software that analyses and filters job applications or evaluates candidates (point (a)). A tool that allocates tasks based on individual behaviour or personal traits, or that monitors and evaluates staff performance and behaviour (point (b)).
- **Point 5, points (b) and (c).** A system that evaluates the creditworthiness of customers who are natural persons or establishes their credit score (for example, to sell in instalments), unless it is used to detect financial fraud; and risk assessment and pricing in life and health insurance.
- **Point 1, point (c).** A system intended for emotion recognition (for example, in call analysis), in so far as its use is permitted. Note: at the workplace, inferring a person's emotions has been prohibited since 2 February 2025, except for medical or safety reasons (Article 5(1), point (f)).
- **Point 3, education.** For academies, driving schools or training centres: systems that decide admission, evaluate learning or monitor tests.
What counts is the purpose for which the provider intends the system ("intended purpose", Article 3, point (12)) and the use made of it. If the company changes the intended purpose of a system that was not high-risk, including a general-purpose one, so that it becomes high-risk, Article 25(1), point (c), considers it the provider of a high-risk system.
The Article 6(3) derogation and the profiling rule
An Annex III system is not considered high-risk 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 (Article 6(3)). This applies where any of four conditions is fulfilled: the system is intended to perform a narrow procedural task; to improve the result of a previously completed human activity; to detect decision-making patterns or deviations without replacing or influencing the previously completed human assessment without proper human review; or to perform a preparatory task to an assessment relevant to the Annex III use cases.
The derogation has a limit: Annex III systems that perform profiling of natural persons "shall always be considered to be high-risk" (Article 6(3), third subparagraph). And the one who relies on it is the provider, which documents its assessment before placing the system on the market and provides it to the authorities on request (Article 6(4)). For the user, the practical question is which classification the provider states and why.
What changes on 2 December 2027 for those who use them
From that date, for Annex III systems, the deployer obligations of Article 26 apply: using them in accordance with the instructions for use, assigning human oversight to people with the necessary competence, training and authority, monitoring their operation and keeping the logs they generate, among others. Three of them concern an SME directly:
- **Article 26(7).** Before putting into service or using a high-risk system at the workplace, the employer informs workers' representatives and the affected workers.
- **Article 26(11).** Whoever uses an Annex III system to make decisions, or assist in making them, related to natural persons informs those persons that they are subject to its use.
- **Article 27.** Bodies governed by public law, private entities providing public services and deployers of systems under point 5, points (b) and (c) (creditworthiness and life and health insurance), perform a fundamental rights impact assessment before deploying them.
For Annex I systems, Article 26 applies from 2 August 2028.
How to review it, system by system
- List the AI systems the company uses, including those built into another program, with their provider and what they are used for.
- Compare each one's use with the eight Annex III areas and record the matching one, or none.
- If it falls within an area, record whether it performs profiling of natural persons and whether any of the four Article 6(3) conditions applies.
- Ask the provider for the classification it states and the documentation of its assessment, and record them.
- Keep the date of the review and repeat it when the system or its use changes; before 2 December 2027, to have the Article 26 obligations ready.
The AI Inventory template comes with this review preloaded: the "Classification (Art. 6 and Annex III)" database has the Article 6 rules and the eight Annex III areas, each row with its summary, the link to EUR-Lex and the date it was checked; and in "AI systems" each system has "Annex III area", "Performs profiling of natural persons (Art. 6(3))", "Article 6(3) derogation condition" and "Classification stated by the provider", which feed its "Indicative level" and the "Possible high risk (indicative)" view: AI Inventory.
Sources (opened on 28 October 2026)
- Regulation (EU) 2024/1689, consolidated text as of 27 July 2026, the current consolidated version given by EUR-Lex (EUR-Lex): https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:02024R1689-20260727 — Articles 3, 5, 6, 25, 26, 27 and 113 and Annexes I and III.
- Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 (OJ L of 24 July 2026): https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32026R1744
Document organisation tool based on official texts; it does not constitute legal advice.