{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,16]],"date-time":"2026-06-16T22:57:13Z","timestamp":1781650633623,"version":"3.54.5"},"reference-count":39,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,4,20]],"date-time":"2026-04-20T00:00:00Z","timestamp":1776643200000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,5,23]],"date-time":"2026-05-23T00:00:00Z","timestamp":1779494400000},"content-version":"vor","delay-in-days":33,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Discov Artif Intell"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>The growing adoption of Artificial Intelligence (AI) in critical sectors raises increasing concerns about regulatory compliance. The European Union Artificial Intelligence Act (EU AI Act) represents the first comprehensive legal framework specifically designed to govern AI systems, classifying them according to their associated risk levels. This paper presents an AI-based compliance checker developed to support the assessment of AI systems under the EU AI Act. The tool is designed to help developers understand legal obligations and identify potential compliance risks through structured, contextualized analysis. We describe the system\u2019s architecture, which combines a generative language model with legal document retrieval techniques to interpret regulatory requirements. The tool was tested on a range of real-world use cases, including algorithmic hiring, medical diagnostics, smart city traffic management, and facial recognition in retail environments. Evaluation was conducted through both automated analysis and expert review. The results confirm the system\u2019s effectiveness in understanding and reconstructing core regulatory obligations, particularly in assessing risk levels and suggesting improvements. While the tool performed strongly in central legal areas, some limitations emerged in handling highly technical or cross-regulatory content. Expert feedback highlighted its utility as a pre-audit tool, emphasizing its depth of legal reasoning, transparency, and contextual adaptability.<\/jats:p>","DOI":"10.1007\/s44163-026-01196-1","type":"journal-article","created":{"date-parts":[[2026,4,20]],"date-time":"2026-04-20T20:59:48Z","timestamp":1776718788000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Evaluating GenAI for automated EU AI Act compliance against human experts"],"prefix":"10.1007","volume":"6","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-3742-0007","authenticated-orcid":false,"given":"Maria Assunta","family":"Cappelli","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"L\u00e9onore","family":"Bozovic","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5048-5251","authenticated-orcid":false,"given":"Giovanna","family":"Di Marzo Serugendo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0254-0794","authenticated-orcid":false,"given":"Jose Luis","family":"Fernandez-Marquez","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,4,20]]},"reference":[{"issue":"1","key":"1196_CR1","doi-asserted-by":"publisher","first-page":"170","DOI":"10.35870\/jtik.v8i1.1416","volume":"8","author":"R Al Rasyid","year":"2024","unstructured":"Al Rasyid R, Ningsih DHU. Penerapan algoritma tf-idf dan cosine similarity untuk query pencarian pada dataset destinasi wisata. Jurnal JTIK (Jurnal Teknologi Informasi dan Komunikasi). 2024;8(1):170\u20138.","journal-title":"Jurnal JTIK (Jurnal Teknologi Informasi dan Komunikasi)"},{"key":"1196_CR2","doi-asserted-by":"publisher","DOI":"10.1016\/j.dibe.2020.100039","volume":"5","author":"R Amor","year":"2021","unstructured":"Amor R, Dimyadi J. The promise of automated compliance checking. Dev Built Environ. 2021;5:100039.","journal-title":"Dev Built Environ"},{"key":"1196_CR3","doi-asserted-by":"publisher","unstructured":"Cappelli MA, Di\u00a0Marzo\u00a0Serugendo G. Reconciling companies and ai regulations initiatives (roccia). Research report, Universit\u00e9 de Gen\u00e8ve. 2023. https:\/\/doi.org\/10.13097\/archive-ouverte\/unige:172701","DOI":"10.13097\/archive-ouverte\/unige:172701"},{"key":"1196_CR4","doi-asserted-by":"publisher","unstructured":"Cappelli MA, Di\u00a0Marzo\u00a0Serugendo G. A semi-automated software model to support AI ethics compliance assessment of an AI system guided by ethical principles of AI. AI Ethics 2025;5:1357\u201380. https:\/\/doi.org\/10.1007\/s43681-024-00480-z","DOI":"10.1007\/s43681-024-00480-z"},{"key":"1196_CR5","doi-asserted-by":"publisher","unstructured":"Cappelli MA, Di\u00a0Marzo\u00a0Serugendo G. Semantic-driven conversational agent for \u201csos rentr\u00e9e\u201d university service. Research report, Universit\u00e9 de Gen\u00e8ve. 2025. https:\/\/doi.org\/10.13097\/archive-ouverte\/unige:187530","DOI":"10.13097\/archive-ouverte\/unige:187530"},{"key":"1196_CR6","unstructured":"Cohen WW, Ravikumar P, Fienberg SE. A comparison of string distance metrics for name-matching tasks. In: Proceedings of the 2003 international conference on information integration on the web, IIWEB\u201903, AAAI Press; 2003. pp. 73\u201378."},{"key":"1196_CR7","doi-asserted-by":"crossref","unstructured":"Deshpande A. Regulatory compliance and ai: Navigating the legal and regulatory challenges of ai in finance. In: 2024 international conference on knowledge engineering and communication systems (ICKECS). 2024. vol. 1; pp.1\u20135.","DOI":"10.1109\/ICKECS61492.2024.10616752"},{"issue":"5","key":"1196_CR8","doi-asserted-by":"publisher","first-page":"C4016001","DOI":"10.1061\/(ASCE)CP.1943-5487.0000572","volume":"30","author":"J Dimyadi","year":"2016","unstructured":"Dimyadi J, Clifton C, Spearpoint M, Amor R. Computerizing regulatory knowledge for building engineering design. J Comput Civ Eng. 2016;30(5):C4016001.","journal-title":"J Comput Civ Eng"},{"key":"1196_CR9","doi-asserted-by":"crossref","unstructured":"Fan Y, Ni J, Niklaus J, Aeschlimann D, Habegger N, Stadelmann J, von Reden A-L, Sennhauser J, Hodel J, Hodel P, et\u00a0al. Lexam: benchmarking legal reasoning on 340 law exams. 2025. arXiv:2505.12864.","DOI":"10.2139\/ssrn.5265144"},{"issue":"1","key":"1196_CR10","doi-asserted-by":"publisher","first-page":"79","DOI":"10.1007\/s10115-017-1142-1","volume":"57","author":"M Hashmi","year":"2018","unstructured":"Hashmi M, Governatori G, Lam H-P, Wynn MT. Are we done with business process compliance: state of the art and challenges ahead. Knowl Inf Syst. 2018;57(1):79\u2013133.","journal-title":"Knowl Inf Syst"},{"key":"1196_CR11","doi-asserted-by":"publisher","unstructured":"He Z, Cao P, Wang C. Jin Z, Chen Y, Xu J, Li H, Jiang X, Liu K, Zhao J. Agentscourt: building judicial decision-making agents with court debate simulation and legal knowledge augmentation. Conference on Empirical Methods in Natural Language Processing. 2024. https:\/\/doi.org\/10.18653\/v1\/2024.findings-emnlp.549","DOI":"10.18653\/v1\/2024.findings-emnlp.549"},{"key":"1196_CR12","doi-asserted-by":"crossref","unstructured":"Imperial JM, Jones MD, Tayyar\u00a0Madabushi H. Standardizing intelligence: aligning generative ai for regulatory and operational compliance. Working paper or preprint. 2025.","DOI":"10.2139\/ssrn.5146161"},{"key":"1196_CR13","doi-asserted-by":"crossref","unstructured":"Ioannidis J, Harper J, Quah MS, Hunter D. Gracenote. ai: legal generative ai for regulatory compliance. In: Proceedings of the third international workshop on artificial intelligence and intelligent assistance for legal professionals in the digital Workplace (LegalAIIA 2023). 2023.","DOI":"10.2139\/ssrn.4494272"},{"issue":"2","key":"1196_CR14","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/1376815.1376819","volume":"2","author":"A Islam","year":"2008","unstructured":"Islam A, Inkpen D. Semantic text similarity using corpus-based word similarity and string similarity. ACM Trans Knowl Discov Data. 2008;2(2):1\u201325.","journal-title":"ACM Trans Knowl Discov Data"},{"key":"1196_CR15","doi-asserted-by":"crossref","unstructured":"Kim J, Min M. From rag to qa-rag: integrating generative ai for pharmaceutical regulatory compliance process. 2024.","DOI":"10.1145\/3672608.3707749"},{"key":"1196_CR16","unstructured":"Lee JK (2011) Building environment rule and analysis (BERA) language and its application for evaluating building circulation and spatial program. PhD thesis, Georgia Institute of Technology."},{"key":"1196_CR17","doi-asserted-by":"crossref","unstructured":"Li H, Chen J, Yang J, Ai Q, Jia W, Liu Y, Lin K, Wu Y, Yuan G, Hu Y, Wang W, Liu Y, Huang M. LegalAgentBench: evaluating LLM agents in legal domain. 2024.","DOI":"10.18653\/v1\/2025.acl-long.116"},{"key":"1196_CR18","doi-asserted-by":"crossref","unstructured":"Libal T. Towards automated gdpr compliance checking. In: Trustworthy AI-integrating learning, optimization and reasoning: first international workshop, TAILOR 2020, Virtual Event, September 4\u20135, 2020, Revised Selected Papers 1. Springer; 2021. pp. 3\u201319.","DOI":"10.1007\/978-3-030-73959-1_1"},{"key":"1196_CR19","unstructured":"Lin C-Y. ROUGE: a package for automatic evaluation of summaries. In: Text summarization branches Out. Barcelona: Spain; 2004. pp. 74\u201381. Association for Computational Linguistics."},{"issue":"2","key":"1196_CR20","doi-asserted-by":"publisher","first-page":"541","DOI":"10.1007\/s10844-023-00782-4","volume":"61","author":"F Lor\u00e8","year":"2023","unstructured":"Lor\u00e8 F, Basile P, Appice A, de Gemmis M, Malerba D, Semeraro G. An ai framework to support decisions on gdpr compliance. J Intell Inf Syst. 2023;61(2):541\u201368.","journal-title":"J Intell Inf Syst"},{"key":"1196_CR21","unstructured":"Magesh V, Surani F, Dahl M, Suzgun M, Manning CD, Ho DE. Hallucination-free? Assessing the reliability of leading ai legal research tools. 2024. arXiv:2405.20362."},{"issue":"2","key":"1196_CR22","first-page":"218","volume":"7","author":"AM Molinero","year":"2006","unstructured":"Molinero AM, Kohun FG, Morris R. Reliability in automated evaluation tools for web accessibility standards compliance. Issues Inf Syst. 2006;7(2):218\u201322.","journal-title":"Issues Inf Syst"},{"key":"1196_CR23","doi-asserted-by":"publisher","first-page":"80","DOI":"10.1016\/j.compag.2011.10.011","volume":"80","author":"R Nikkil\u00e4","year":"2012","unstructured":"Nikkil\u00e4 R, Wiebensohn J, Nash E, Seilonen I, Koskinen K. A service infrastructure for the representation, discovery, distribution and evaluation of agricultural production standards for automated compliance control. Comput Electron Agric. 2012;80:80\u20138.","journal-title":"Comput Electron Agric"},{"key":"1196_CR24","unstructured":"Niwattanakul S, Singthongchai J, Naenudorn E, Wanapu S. Using of Jaccard coefficient for keywords similarity. In: IMECS. 2013."},{"key":"1196_CR25","doi-asserted-by":"crossref","unstructured":"Preidel C, Borrmann A. Automated code compliance checking based on a visual language and building information modeling. In ISARC. Proceedings of the international symposium on automation and robotics in construction. IAARC Publications; 2015. vol. 32, pp. 1.","DOI":"10.22260\/ISARC2015\/0033"},{"key":"1196_CR26","unstructured":"Reimers N. all-MiniLM-L6-v2. 2021. https:\/\/huggingface.co\/sentence-transformers\/all-MiniLM-L6-v2. Accessed 30 June 2025."},{"key":"1196_CR27","doi-asserted-by":"crossref","unstructured":"Reimers N, Gurevych I. Sentence-BERT: sentence embeddings using siamese BERT-networks. In: Inui K, Jiang J, Ng V, Wan X. (eds.), Proceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP), Hong Kong: China; 2019. pp. 3982\u20133992. Association for Computational Linguistics.","DOI":"10.18653\/v1\/D19-1410"},{"issue":"2","key":"1196_CR28","doi-asserted-by":"publisher","first-page":"505","DOI":"10.1007\/s10506-023-09360-z","volume":"32","author":"L Robaldo","year":"2024","unstructured":"Robaldo L, Batsakis S, Calegari R, Calimeri F, Fujita M, Governatori G, et al. Compliance checking on first-order knowledge with conflicting and compensatory norms: a comparison among currently available technologies. Artif Intell Law. 2024;32(2):505\u201355.","journal-title":"Artif Intell Law"},{"key":"1196_CR29","unstructured":"Sai C, Sadiq S, Han L, Demartini G, Rinderle-Ma S. Identification of regulatory requirements relevant to business processes: a comparative study on generative AI, embedding-based ranking, crowd and expert-driven methods. 2024. arXiv:2401.02986."},{"key":"1196_CR30","doi-asserted-by":"publisher","first-page":"69","DOI":"10.1016\/j.autcon.2015.03.003","volume":"53","author":"W Solihin","year":"2015","unstructured":"Solihin W, Eastman C. Classification of rules for automated bim rule checking development. Autom Constr. 2015;53:69\u201382.","journal-title":"Autom Constr"},{"issue":"2","key":"1196_CR31","doi-asserted-by":"publisher","first-page":"697","DOI":"10.30574\/wjarr.2024.23.2.2364","volume":"23","author":"ID Ssetimba","year":"2024","unstructured":"Ssetimba ID, Kato J, Pinyi EO, Twineamatsiko E, Nakayenga HN, Muhangi E. Advancing electronic communication compliance and fraud detection through machine learning, nlp and generative ai: a pathway to enhanced cybersecurity and regulatory adherence. World J Adv Res Rev. 2024;23(2):697\u2013707.","journal-title":"World J Adv Res Rev"},{"issue":"06","key":"1196_CR32","doi-asserted-by":"publisher","first-page":"9019","DOI":"10.15680\/IJIRSET.2023.1206157","volume":"12","author":"S Tiwari","year":"2023","unstructured":"Tiwari S, Sahu R, Shripal S, Rajpoot SS. Innovative framework for context-based recommendation using tf-idf and cosine similarity. Int J Innov Res Sci Eng Technol. 2023;12(06):9019\u201323.","journal-title":"Int J Innov Res Sci Eng Technol"},{"key":"1196_CR33","unstructured":"Wang W, Wei F, Dong L, Bao H, Yang N, Zhou M. Minilm: deep self-attention distillation for task-agnostic compression of pre-trained transformers. In: Proceedings of the 34th international conference on neural information processing systems, NIPS \u201920, Red Hook, NY, USA. Curran Associates Inc. 2020."},{"key":"1196_CR34","unstructured":"Wikipedia Contributors. Jaccard index\u2014wikipedia, the free encyclopedia; 2023a. https:\/\/en.wikipedia.org\/wiki\/Jaccard_index. Accessed 07 Jan 2025."},{"key":"1196_CR35","unstructured":"Wikipedia Contributors. Longest common subsequence\u2014wikipedia, the free encyclopedia; 2023b. https:\/\/en.wikipedia.org\/wiki\/Longest_common_subsequence. Accessed 07 Jan 2025."},{"key":"1196_CR36","doi-asserted-by":"crossref","unstructured":"Zhang J, El-Gohary N. Extraction of construction regulatory requirements from textual documents using natural language processing techniques. 2012; pp. 453\u2013460. American Society of Civil Engineers.","DOI":"10.1061\/9780784412343.0057"},{"key":"1196_CR37","doi-asserted-by":"crossref","unstructured":"Zhang J, El-Gohary NM. Information transformation and automated reasoning for automated compliance checking in construction. 2013; pp. 701\u2013708. American Society of Civil Engineers.","DOI":"10.1061\/9780784413029.088"},{"key":"1196_CR38","unstructured":"Zhang T, Kishore V, Wu F, Weinberger KQ, Artzi Y. Bertscore: evaluating text generation with bert. 2020."},{"key":"1196_CR39","unstructured":"Zhu L, Yang L, Li C, Hu S, Liu L, Yin B. Legilm: a fine-tuned legal language model for data compliance. 2024. arXiv:2409.13721."}],"container-title":["Discover Artificial Intelligence"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s44163-026-01196-1","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44163-026-01196-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s44163-026-01196-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,16]],"date-time":"2026-06-16T22:26:41Z","timestamp":1781648801000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s44163-026-01196-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,20]]},"references-count":39,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["1196"],"URL":"https:\/\/doi.org\/10.1007\/s44163-026-01196-1","relation":{},"ISSN":["2731-0809"],"issn-type":[{"value":"2731-0809","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,20]]},"assertion":[{"value":"11 September 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"20 March 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"20 April 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare that they have no conflict of interest.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}},{"value":"Not applicable.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent to publish"}},{"value":"Not applicable.","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent to participate"}},{"value":"The study underwent an institutional ethical self-assessment procedure in accordance with the current guidelines of the University of Geneva.","order":5,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics approval"}}],"article-number":"458"}}