To Be High-Risk, or Not To Be—Semantic Specifications and Implications of the AI Act’s High-Risk AI Applications and Harmonised Standards
Open accessThe DOI matches its Crossref record; 17 of the 17 references it cites were found in a registry or an index.
Abstract
The EU’s proposed AI Act sets out a risk-based regulatory framework to govern the potential harms emanating from use of AI systems. Within the AI Act’s hierarchy of risks, the AI systems that are likely to incur “high-risk” to health, safety, and fundamental rights are subject to the majority of the Act’s provisions. To include uses of AI where fundamental rights are at stake, Annex III of the Act provides a list of applications wherein the conditions that shape high-risk AI are described. For high-risk AI systems, the AI Act places obligations on providers and users regarding use of AI systems and keeping appropriate documentation through the use of harmonised standards. In this paper, we analyse the clauses defining the criteria for high-risk AI in Annex III to simplify identification of potential high-risk uses of AI by making explicit the “core concepts” whose combination makes them high-risk. We use these core concepts to develop an open vocabulary for AI risks (VAIR) to represent and assist with AI risk assessments in a form that supports automation and integration. VAIR is intended to assist with identification and documentation of risks by providing a common vocabulary that facilitates knowledge sharing and interoperability between actors in the AI value chain. Given that the AI Act relies on harmonised standards for much of its compliance and enforcement regarding high-risk AI systems, we explore the implications of current international standardisation activities undertaken by ISO and emphasise the necessity of better risk and impact knowledge bases such as VAIR that can be integrated with audits and investigations to simplify the AI Act’s application.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“There is currently also a gap between the current state of AI standardization at the international level and the desired state of EU-wide harmonized standards required for compliance with the AI Act.”
From A requirements model for AI algorithms in functional safety-critical systems with an explainable self-enforcing network from a developer perspective (2024) · cites this paper as [15]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
1 retracted
Oprea, A., & Vassilev, A. T. (2023) · doi:10.6028/nist.ai.100-2e2023.ipd · Cited by 46
RetractedVeale, M., & Zuiderveen Borgesius, F. (2021) · Computer Law Review International · doi:10.9785/cri-2021-220402 · Cited by 805 · cited 2×
“40), AI providers can achieve compliance with the requirements through conformance to harmonised standards, without undergoing the costly and time-consuming process of requirements interpretation [24].”
BackgroundTaxonomy of Risks posed by Language Models
Weidinger, L., Uesato, J., Rauh, M., Griffin, C., Huang, P.-S., Mellor, J., Glaese, A., Cheng, M., Balle, B., Kasirzadeh, A., et al. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533088 · Cited by 764 · cited 1×
“Examples of these are: Weidinger et al.’s taxonomy of ethical and social risks of language models [25], the open loop’s taxonomy of potential harms associated with machine learning applications and automated decision-making systems [5], NIST’s taxonomy of adversarial machine learning [15] and…”
BackgroundTowards a standard for identifying and managing bias in artificial intelligence
Schwartz, R., Vassilev, A., Greene, K., Perine, L., Burt, A., & Hall, P. (2022) · doi:10.6028/nist.sp.1270 · Cited by 548 · cited 1×
“…learning applications and automated decision-making systems [5], NIST’s taxonomy of adversarial machine learning [15] and categories of AI Bias [21], Steimers and Schneider’s work on creating a taxonomy of risk sources that impact AI trustworthiness [23], and Roselli et al.’s work on…”
Roselli, D., Matthews, J., & Talagala, N. (2019) · Companion Proceedings of The 2019 World Wide Web Conference · doi:10.1145/3308560.3317590 · Cited by 237 · cited 1×
“’s work on classification of AI bias [19].”
BackgroundPreventing repeated real world AI failures by cataloging incidents: the AI incident database
McGregor, S. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i17.17817 · Cited by 156
Ebers, M., Hoch, V. R. S., Rosenkranz, F., Ruschemeier, H., & Steinrötter, B. (2021) · J · doi:10.3390/j4040043 · Cited by 153 · cited 2×
“This limitation in adding new areas is criticised in [7] and [22], where the authors highlight the necessity of extending the high-risk areas.”
BackgroundSmuha, N. A., Rengers, E., Harkens, A., Li, W., MacLaren, J., Piselli, R., & Yeung, K. (2021) · doi:10.2139/ssrn.3899991 · Cited by 153
Steimers, A., & Schneider, M. (2022) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph19063641 · Cited by 151 · cited 1×
“…and automated decision-making systems [5], NIST’s taxonomy of adversarial machine learning [15] and categories of AI Bias [21], Steimers and Schneider’s work on creating a taxonomy of risk sources that impact AI trustworthiness [23], and Roselli et al.’s work on classification of AI bias [19].”
BackgroundOECD Framework for the Classification of AI systems
(2022) · OECD Digital Economy Papers · doi:10.1787/cb6d9eca-en · Cited by 112 · cited 1×
“Developing a common framework for reporting AI incidents is on the OECD’s agenda for future work [14].”
BackgroundInformation Control and Public Support for Social Credit Systems in China
Xu, X., Kostka, G., & Cao, X. (2021) · The Journal of Politics · doi:10.1086/718358 · Cited by 108 · cited 1×
“5( 1)(c)) while an implementation of this is being used in China [26].”
MethodComing to Terms with FAIR Ontologies
Poveda‐Villalón, M., Espinoza-Arias, P., Garijo, D., & Oscar Corcho, B. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-61244-3_18 · Cited by 52 · cited 1×
“In creation of VAIR, we considered rules suggested by Poveda-Villalón et al. [18] to ensure its FAIRness.”
MethodGolpayegani, D., Pandit, H. J., & Lewis, D. (2022) · Studies on the semantic web · doi:10.3233/ssw220008 · Cited by 50 · cited 2×
“The AI risk ontology (AIRO) [11] is an ontology for modelling AI systems and their associated risks.”
BackgroundMethodSelf-citationA taxonomic system for failure cause analysis of open source AI incidentsPreprint
Pittaras, N., & McGregor, S. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2211.07280 · Cited by 28 · cited 1×
“The creation of the taxonomy followed a bottom-up approach where the taxonomy is populated through incident annotation [17].”
MethodA Semantic Specification for Data Protection Impact Assessments (DPIA)
Pandit, H. J. (2022) · Studies on the semantic web · doi:10.3233/ssw220007 · Cited by 11 · cited 1×
“Using VAIR alongside existing vocabularies that concern risk, such as the Data Privacy Vocabulary (DPV)9, facilitates integration of existing risk management and impact assessment practices when dealing with multiple EU regulations and conducting shared impact assessments [16].”
BackgroundSelf-citationTempting the Fate of the furious: cyber security and autonomous cars
McLachlan, S., Schäfer, B., Dube, K., Kyrimi, E., & Fenton, N. (2022) · International Review of Law Computers & Technology · doi:10.1080/13600869.2022.2060466 · Cited by 8 · cited 1×
“Although compliance with these standards is not enforced [13], when a high-risk AI conforms to the harmonised 4 shows the alignment of the AI Act’s risk management system steps (Art.”
BackgroundNewsFinder: Automating an AI News Service
Eckroth, J., Dong, L., Smith, R. G., & Buchanan, B. G. (2012) · AI Magazine · doi:10.1609/aimag.v33i2.2406 · Cited by 6 · cited 1×
“Discovery, categorisation (determining the main focus), and summarisation of AI news featured in AITopics are automated [8].”
Background
Research works printed but not linked: 3
- Jerome De Cooman. 2022. Humpty dumpty and high-risk AI systems: the ratione materiae dimension of the proposal for an EU artificial intelligence act. Mkt. & Competition L. Rev. 6 (2022), 49.
- European Commission, Joint Research Centre, Stefano Nativi, and Sarah De Nigris. 2021. AI Watch: AI standardisation landscape state of play and link to the EC proposal for an AI regulatory framework. Technical Report. https://data.europa.eu/doi/10. 2760/376602 · doi:10.2760/376602
- European Commission, Joint Research Centre, Josep Soler Garrido, Songül Tolan, Isabelle Hupont Torres, David Fernandez Llorca, Vicky Charisi, Emilia Gomez Gutierrez, Henrik Junklewitz, Ronan Hamon, Delia Fano Yela, and Cecilia Panigutti. 2023. AI Watch: Artificial Intelligence Standardisation Landscape Update. Analysis of IEEE standards in the context of the European AI Regulation. Technical Report. Luxembourg (Luxembourg). https://data.europa.eu/doi/10.2760/131984 · doi:10.2760/131984
Cited by
Lendvai, G. F., & Gosztonyi, G. (2025) · Laws · doi:10.3390/laws14030041 · Cited by 63
Data Privacy Vocabulary (DPV) – Version 2.0
Pandit, H. J., Esteves, B. G. C., Krog, G. P., Ryan, P., Golpayegani, D., & Flake, J. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-77847-6_10 · Cited by 42
Self-citationBlockchain-Enabled Federated Learning in Healthcare: Survey and State-of-the-Art
Nezhadsistani, N., Moayedian, N. S., & Stiller, B. (2025) · IEEE Access · doi:10.1109/access.2025.3587345 · Cited by 47
“Such systems will be subject to strict regulations, such as risk management plans, human intervention, and clear declaration of data use [143].”
Golpayegani, D., Hupont, I., Panigutti, C., Pandit, H. J., Schade, S., O’Sullivan, D., & Lewis, D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-68024-3_3 · Cited by 58
Self-citationDu, J., Cai, H., & Jin, X. (2024) · Sustainability · doi:10.3390/su16219315 · Cited by 21
Metadata
- Authors
- Golpayegani, D., Pandit, H. J., & Lewis, D.
- Affiliations
- Trinity College Dublin (IE)
- Dublin City University (IE)
- ADAPT Centre, Trinity College Dublin, Ireland
- Venue
- 2023 ACM Conference on Fairness Accountability and Transparency
- Publisher
- ACM
- Published
- 2023-06-12
- Pages
- 905-915
- OpenAlex
- W4380319304
- Cited by
- 42
- References
- 15
- Citations by year
- 2026: 15 · 2025: 13 · 2024: 14
- FWCI
- 116.25
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- ACM Conference on Fairness, Accountability, and Transparency (FAccT) (publishedVersion, cc-by)
- Dublin City University Open Access Institutional Repository (Dublin City University) (submittedVersion, cc-by)
- Trinity's Access to Research Output (TARA) (Trinity College Dublin) (submittedVersion, cc-by-nc-sa)
- Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, cc-by)
- Zenodo (CERN European Organization for Nuclear Research) (cc-by)
- Funders
- European Commission (813497)
- Science Foundation Ireland (SFI) (13/RC/2106_P2)
- Topics
- Social Sciences › Law › Digitalization, Law, and Regulation
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Engineering › Safety, Risk, Reliability and Quality › Safety Systems Engineering in Autonomy
- Keywords
- EU AI Act, high-risk AI systems, risk-based regulation, harmonised standards, AI risk assessment, semantic specification, fundamental rights, risk documentation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.