1. Introduction

Artificial Intelligence (AI) has become synonymous with technological advancement and is one of the most significant applications of the data-driven economy¹ . At its core, AI is defined² as “a technology that enables computers to perform tasks requiring intelligence when performed by humans” ³.

Artificial Intelligence (AI) and Machine Learning (ML) have emerged as transformative forces in various sectors, revolutionising processes and enhancing efficiency. More precisely in the field of justice, digital advancements facilitated by AI can be categorised into three key areas: a) Improving information accessibility: AI tools provide parties with information on judicial procedures and available options related thereto. They also provide support for the judicial system in general⁴ through case management and automated classification systems, b) Reshaping traditional methods of interpersonal communications: Videoconferencing and electronic messaging are now common in the judicial process, and c) Redefining judicial workflows: Judges rely on electronic⁵ and legal databases⁶. In the future, AI may further enhance judicial efficiency by predicting case outcomes.

Public sector auditing, a domain traditionally reliant on rigorous analysis and detailed oversight, is now poised to reap the benefits of these advanced technologies. For the Hellenic Court of Audit (HCA), which is both the Supreme Financial Court and the Supreme Audit Institution in Greece, the integration of AI and ML offers opportunities to modernise auditing methodologies, improve efficiency, and strengthen accountability. However, it also presents challenges, particularly in ethics and accountability, legal and regulatory barriers, as well as technical and resource limitations.

2. Opportunities for the Hellenic Court of Audit (HCA)

AI inherently has the potential to transform public sector auditing by automating time-consuming tasks and analysing complex datasets with unparalleled speed and precision. These capabilities are particularly valuable for: a) Document analysis: Natural language processing (NLP) can assist in reviewing extensive documents, extracting critical information and highlighting inconsistencies⁷ , b) Risk assessment⁸ : AI systems can rapidly identify anomalies, patterns, and trends, enabling auditors to focus on high-risk areas, and c) Fraud detection⁹: AI algorithms can flag suspicious transactions or activities, reducing the likelihood of undetected fraud.

Moreover, as the adoption of AI aligns with global trends in public administration, where digital transformation has become essential, AI tools can be used to monitor the reliability of auditees’ information systems, develop predictive models for financial and operational risks, and automate routine tasks, so as to enhance the overall effectiveness of the HCA audit procedures¹⁰.

Finally, AI-powered analytics, by offering data-driven insights, can contribute to identifying patterns in financial and administrative processes, as well as linking findings to the relevant legal framework, with the view to strengthening decision-making.

In the light of the above, the HCA can leverage AI to enhance the efficiency and accuracy of public audit, modernise auditing practices and support strategic decision-making.

3. Innovations implemented by a SAI with jurisdictional competences

The COVID-19 pandemic highlighted the resilience of justice systems and the urgency of digitalisation. Amendments across legal frameworks accelerated the adoption of digital and AI-driven solutions, as exemplified by Greece's judicial reforms.

In this context, recognising the importance of integrating digital tools into the judicial sector, Greece has embarked on its most extensive programme of digital reforms to date, funded by national resources, the EU, and the Recovery and Resilience Facility (RRF). This initiative aligns with the national digital strategy outlined by Law 4727/2020 and the Digital Transformation Bible¹¹. A key component of this strategy, introduced in January 2023, is the development of an Integrated Information System (IIS) specifically for the HCA. This System aims to modernise the HCA’s operations in both audit and judicial fields.

The legal background of this transformative process lies in the HCA Organic Law 4820/2021, which, among others, amended the Court’s code of procedure. By the new legal provisions, judicial decisions, reports, summons, and other court documents are, in principle, electronically notified to the parties’ email addresses¹². Such notification shall be deemed complete once the addressee returns the electronic acknowledgment of receipt to the sender. In the event of failure to return the electronic transmission within 5 days, it shall be presumed, for all intents and purposes, that the notification took place 10 days following the date of the electronic transmission. Furthermore, special provision is made¹³ for the mandatory electronic filing of case-file documents, provided that they are authenticated with a qualified electronic signature, while all associated court fees are paid electronically. Remote access to electronic case-file is granted to the parties and the State Prosecutor General at the HCA through secure connection and credentials¹⁴. Along the same lines, provision is made for participation in court hearings through remote connection technologies (telematics applications)¹⁵.

Moreover, a specific provision was introduced¹⁶ for the development of ML software related to procedural documents with a view to their thematic classification and the identification of relevant case law and legislation¹⁷, so as to form the legal reasoning of the final judgment. At a later stage, the software will be able to identify all the critical data so as to form the factual basis of the judgment. To facilitate the above capacities, a technical format for procedural documents’ formatting will be adopted by the Court’s Plenum, while automatic judgment anonymisation is supported.

In addition to the above rules, which primarily concern the exercise of the HCA’s jurisdictional powers, reform initiatives have been launched regarding audit procedures. These include provisions for equal digital technology both for checking the reliability of the auditees’ information systems and for carrying out audits by the HCA, by developing or procuring AI software capable of detecting problematic accounts, transactions, or actions¹⁸. For this purpose, new audit procedures, including AI-based technologies, are introduced, staff upgrades, including the creation of specialised posts, are established, and the possibility to engage external contractors, when necessary, is provided for¹⁹.

Undeniably, however, in the field of audit, the big step taken is the introduction of the aforementioned new Integrated Information System of the HCA. The system’s key features include: a) Automated case management: Centralised monitoring of audit -as well as judicial- cases, b) Digital workflows: Full digitalisation of audit, administrative, and judicial procedures, c) Data insights: Tools for generating statistical reports and specialised analytics, and d) Videoconferencing: Advanced support for remote hearings and deliberations.

More specifically, the new system is designed to support the various types of audits conducted by the HCA, in accordance with its constitutional mandate as specified by law, to monitor, manage, and document audit cases. The system aims to facilitate the standardisation and organisation of audits, record the procedures for each audit type, track progress, promptly evaluate audit results, and oversee the timely and accurate execution of the HCA Annual Audit Programme. Through interoperability with external judicial and administrative information systems, it will enable seamless digital submission and review of documents and accounts, while secure access will be granted to authorised parties. Additionally, it will allow auditors to use standardised digital tools, such as questionnaires, planning notes, audit programmes, reports, and manuals, alongside technical tools for data analysis. The system will also ensure the supervision and quality assessment of audits at every stage, allow the extraction of specific reports and indicators, and manage digital audit files and a centralised digital audit data repository.

It is understood from the above that the use of digital and AI means, in both judicial and audit fields, is, and will prove to be, a valuable administrative support for the HCA and its environment, by ensuring transparency, efficiency, reduction of bureaucratic procedures, and faster handling of judicial and audit cases as machines work indefinitely²⁰. The potential that the use of AI offers to the HCA, through techniques of natural language processing, machine learning, and neural networks, ensures effectiveness, accountability and equality, and, thus, contributes to the successful fulfilment of its institutional role.

4. Challenges in AI integration and ML implementation at the HCA

While the benefits are clear, implementing AI and ML in public sector auditing comes with significant challenges. It should not be forgotten that the use of AI poses serious risks, related to fundamental rights, discrimination, quality and security, transparency, impartiality and fairness, and uncontrolled use²¹. Undeniably, the success of AI in auditing relies heavily on the quality and integrity of the data it processes, as their lack can lead to Bias and discrimination²² and raise Accountability Issues²³, ²⁴. The HCA can thoroughly address these issues, by correspondingly eliminating any algorithms trained on biased datasets and determining responsibility for errors or eventual unethical decisions.

As to the challenge of Legal and Regulatory barriers, the HCA must ensure continuous compliance with stringent European and national data protection²⁵ and AI ethics laws²⁶, to ultimately safeguard the responsible use of the newly introduced digital tools.²⁷

Furthermore, it is common knowledge that implementing AI technologies requires significant investments in infrastructure²⁸, skilled personnel, and ongoing maintenance. To address the challenge of Technical and Resource limitations, the HCA must bridge gaps in expertise and eventual future budget constraints, potentially through partnerships with academic and other public entities.

Last but not least, AI can augment but not replace human judgment. Judicial and audit reasoning often require contextual understanding involving not only economic but also social considerations that AI cannot replicate. A judge’s—or auditor's—role extends beyond intelligence to include consciousness, self-awareness, and moral reasoning, which are vital for interpreting ambiguous legal or ethical issues.

5. Identifying and following best practices

The HCA can draw valuable insights from international practices. For instance, Norway's Public Sector AI Strategy emphasises a coordinated, ethical approach to AI adoption, focusing on transparency and accountability²⁹. The European Court of Auditors (ECA) has integrated AI to enhance audit processes and develop methodologies for auditing AI systems themselves³⁰. The ECA's focus on training, inter-institutional collaboration, and ethical AI use can serve as a blueprint for the HCA for realising its vision in the areas of justice and audit.

Taking advantage of established good practices and experiences, such as the above-mentioned ones, and in order to enhance the integration of AI and ML into its auditing framework, the HCA should first and for most consider the following steps³¹: a) Develop a clear and comprehensive AI Strategy with a view to enhancing public trust in the HCA’s function, b) Invest in ongoing training and capacity building by offering interested and qualified staff the opportunities to upskill, c) Foster collaboration with peer Supreme Audit Institutions (SAIs) and academia to share resources, knowledge, and best practices, and d) Enhance cybersecurity by strengthening defences against cyber threats to protect sensitive data and, thus, safeguard data integrity or security.³²

6. Conclusion

For the HCA, embracing AI and ML represents an opportunity to modernise its practices and reinforce its role in ensuring proper administration of justice, accountability, and good governance. By addressing the challenges thoughtfully and learning from international best practices and experiences, the HCA can optimise its work. The digital, under human oversight, transformation of judicial and audit processes represents a pivotal advancement in enhancing efficiency, transparency, and public trust. This modernisation will undoubtedly consolidate the institutional role of the HCA, in both fields of justice and audit, while fostering greater responsiveness to the State’s and citizens’ needs.

¹JTC Resource Bulletin Introduction to AI to Courts, Version 1.0 Adopted 27 March 2020, p.1.

²The exact definition of AI has been the subject of numerous discussions.

³Mitrou L., Tassis S., Kosti I., Vorras A., Karakatzounis V, Can the algorithm be moral, fair, transparent, able to judge and administer? (University Publications of Crete, Heraklion 2023, p. 15).

⁴This is the USA “e-Discovery” system, which, through machine learning, extracts the most relevant parts of the file documents, on the basis of a parties’ agreement on the search terms and coding used, certified by the Judge.

⁵For example, in Estonia, in 2005, the so called “e-file system” was developed, consisting of an electronic platform with safeguards, to which citizens enter and register their case.

⁶Contini F., Artificial Intelligence and the Transformation of Humans, law and Technology, Interaction in Judicial Proceedings, Law, Technology and Humans, Vol 2 (1) 2020, p. 2.

⁷Mahadevkar, S.V., Patil, S., Kotecha, K. et al. Exploring AI-driven approaches for unstructured document analysis and future horizons. J Big Data 11, pp 92 (2024). https://doi.org/10.1186/s40537-024-00948-z

⁸Onwubuariri E., Adelakun B., Olaiya P-O, Ziorklui J. (2024). AI-Driven risk assessment: Revolutionizing audit planning and execution. Finance & Accounting Research Journal 6, pp 1069-1090.

⁹Artificial Intelligence in the Public Sector: Summary Note (English). Equitable Growth, Finance and Institutions Note Washington, D.C.

¹⁰Olubusola Odeyemi, Kehinde Feranmi Awonuga, Noluthando Zamanjomane Mhlongo, Ndubuisi Leonard Ndubuisi, Funmilola Olatundun Olatoye and Andrew Ifesinachi Daraojimba, The role of AI in transforming auditing practices: A global perspective review,

¹¹https://digitalstrategy.gov.gr/en/

¹²Law 4700/2020 art. 340, as amended by Law 4820/2021 art. 170.

¹³Ibid art. 342 of the same Law, as replaced by Law 4820/2021 art. 172.

¹⁴Ibid art. 343 par. 1, as replaced by Law 4820/2021 art. 173.

¹⁵Ibid art. 343 par. 2 as inserted by Law 4963/2022 art. 48.

¹⁶Ibid art. 346, as replaced by Law 4820/2021 art. 176.

¹⁷Page 122 of the explanatory memorandum to Law 4820/2021, according to which the established AI tools will quickly and more easily categorise cases, identify applicable laws and precedents, and assist judges by integrating relevant materials into case files.

¹⁸Law 4820/2021 art. 89 (3).

¹⁹Law 4820/2021 art. 16 (2) and page 92 of the explanatory memorandum.

²⁰Sourdin T., supra, p. 1120.

²¹Cf. European Commission for the Efficiency of Justice (CEPEJ), European Charter of Ethics on the use of artificial intelligence in judicial systems and their environment, 3-4 December 2018.

²²Varona, D., Suárez, J. (2022) Discrimination, Bias, Fairness, and Trustworthy AI Applied Sciences

²³Sacramed, M., T. (2024) Reviewing the Philippines Legal Landscape of Artificial Intelligence (AI) in Business: Addressing Bias, Explainability, and Algorithmic Accountability International Journal of Research and Innovation in Social Science

²⁴Islam, M. (2024) Ethical Considerations in AI: Navigating the Complexities of Bias and Accountability Journal of Artificial Intelligence General science (JAIGS) ISSN:3006-4023

²⁵Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (Text with EEA relevance), OJ L 119, 4.5.2016, p. 1–88 https://eur-lex.europa.eu/eli/reg/2016/679/oj

²⁶See Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence, supra footnote 3. For more information see Kouroupis, K. (2023) The AI Act in light of the EU Digital Agenda: A critical approach Journal of Data Protection & Privacy, Pehlivan, C., N. (2024) The EU Artificial Intelligence (AI) Act: An Introduction Global Privacy Law Review.

²⁷Ortega, E., Tran, M., Bandeen, G. (2023) AI Digital Tool Product Lifecycle Governance Framework through Ethics and Compliance by Design† 2023 IEEE Conference on Artificial Intelligence (CAI), 353-356

²⁸Onwubuariri, Ebere & Adelakun, Beatrice & Olaiya, Omolara & Ziorklui, Joseph. (2024), supra, p. 364

²⁹OECD (2024), Artificial intelligence in the public sector, in The Digital Transformation of Norway's Public Sector, OECD Publishing, Paris, https://doi.org/10.1787/869f83c1-en.

³⁰European Court of Auditors, Artificial Intelligence initial strategy and deployment roadmap – 2024-2025, Publications Office of the European Union, 2024, https://data.europa.eu/doi/10.2865/317443

³¹Dr. Sutthi Suntharanurak, Leveraging AI in Performance Auditing: A Feasibility Study for the State Audit Office of Thailand, https://intosaijournal.org/journal-entry/leveraging-ai-in-performance--auditing-a-feasibility-study-for- the-state-audit-office-of-thailand/

³²Grover, S., Broll, B., Babb, D. (2023) Cybersecurity Education in the Age of AI: Integrating AI Learning into Cybersecurity High School Curricula Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1, Cattell, S., Ghosh, A., Kaffee, L. (2024) Coordinated Flaw Disclosure for AI: Beyond Security Vulnerabilities Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society.