Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Engineering the Law-Machine Learning Translation Problem: Developing Legally Aligned Models

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proposes a five-stage framework for developing machine learning models that respect multiple legal obligations at once, by translating each obligation into operational choices and mapping the resulting trade-offs before…

desk verdict A clearly written methodological framework paper that names a real gap in legal-ML development, but its central claim about facilitating legal justification rests on unvalidated proxy metrics and an illustrative, not empirical, case study. read the letter →

arxiv 2504.16969 v1 pith:XTRPKQSH submitted 2025-04-23 cs.CY cs.LG

Mathias Hanson , Gregory Lewkowicz , Sam Verboven This is my paper · ORCID
classification cs.CYcs.LG
keywords artificialintelligencemachinelearningregulatorycompliancelegalriskmanagementAIgovernancehuman-centricanti-moneylaundering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Machine learning models cannot encode legal obligations directly, because model behavior is learned from data and the law is abstract, open to interpretation, and enforced by authorities. The paper's answer is a five-stage design process in which lawyers turn applicable legal obligations into concrete legal requirements, an interdisciplinary team translates those requirements into several defensible technical operationalizations and evaluation metrics, models are trained from combinations of those choices, and the resulting trade-offs between legal proxy metrics and predictive performance are mapped before one model is selected. The selection is accompanied by a documented, proportionality-based legal justification. The authors illustrate the process with an anti-money laundering case study where a random-forest model is chosen by balancing data minimization, re-identification risk, non-discrimination disparity, explainability, and recall. The value of the contribution is that current software engineering methods either review code line by line, which is impossible for learned models, or develop ML models without confronting legal ambiguity and interactions between legal obligations.

What carries the argument

The central mechanism is the five-stage framework: (1) identification of applicable legal obligations and their translation into legal requirements; (2) translation of each requirement into one or more technical operationalizations together with an evaluation metric or heuristic; (3) formation of operationalization sets, each a full combination of one operationalization per requirement, and training a model on each compatible set; (4) trade-off mapping, in which proxy metric values and predictive performance are tabulated per model; and (5) model selection and legal justification, in which the team chooses a model that is adequate across all dimensions and documents why the trade-offs are legally acceptable. The load-bearing idea is that a legal obligation has no direct implementation in an ML model; it only exists through chosen operationalizations and proxy metrics, so the framework makes those choices explicit and comparable.

What would settle it

Look for a real deployment where a court or regulator later issues a binding decision on whether the model violated the law. If the selected model had good scores on the framework's chosen measurements, such as low group disparity, minimized data use, and low re-identification risk, yet the authority still finds a violation, or if a model with worse scores is found lawful, then the central assumption that the measurements track the law is wrong.

Watch

Extended reading notes

Core claim

The paper's central claim is that legal obligations can be incorporated into ML development through an indirect, staged translation process: law to legal requirements, legal requirements to multiple operationalizations, operationalizations to a portfolio of trained models, and then trade-off mapping to a justified selection. Because laws are abstract and permit multiple valid interpretations, no single metric can prove compliance; instead, organizations should generate a portfolio of models from combinations of defensible operationalizations, evaluate them with legally chosen proxy metrics, and use the resulting trade-off map to select a model and document a proportionality-based legal justification. The anti-money laundering case study demonstrates how this produces a concrete choice that balances data minimization, re-identification risk, non-discrimination disparity, explainability, and recall.

Load-bearing premise

The framework assumes that a legal obligation can be meaningfully captured by the concrete choices and measurement numbers used to represent it, so that the trade-offs seen between those numbers are a trustworthy picture of trade-offs between the legal obligations themselves.

Editorial extensions

If this is right

  • A documented legal justification becomes a built-in output of model development rather than a post-hoc add-on, giving organizations an audit trail for regulators.
  • The framework turns legal ambiguity into an explicit portfolio: instead of searching for one compliant model, organizations compare several defensible models and choose with full information about the trade-offs.
  • Because trade-offs between operationalizations are a priori unknown, legal alignment cannot be reasoned about in the abstract; it has to be measured on trained models.
  • The same five stages transfer to other regulated settings such as finance, healthcare, and public administration by swapping the legal analysis and the proxy metrics.
  • Legal norms evolve over time, so the trade-off maps and justifications should be revisited after deployment, as the paper notes that metrics are time-limited for legal assessment.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The biggest practical risk is not in the engineering but in the legal semantics: if a chosen proxy metric diverges from what an authority later treats as a violation, the trade-off maps will look reassuring while being wrong; an empirical validation loop against actual regulatory outcomes would address this.
  • The framework implicitly treats legal justification as a design artifact; a natural extension is a living document that must be updated before every deployment or retraining, not just at first selection.
  • Because the case study table uses illustrative values, the framework's practical value depends on whether real-world trade-off maps are stable and reproducible across datasets; running the same five stages on public datasets would test that.
  • A regulator could invert the framework: collect many organizations' trade-off maps to identify recurring legal conflicts and target rule-making where proxies for different obligations consistently collide.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper argues that current software/requirements engineering methods are inadequate for ML models subject to multiple, uncertain legal obligations, because legal obligations cannot be directly encoded in source code and must instead be operationalized through reductive proxy metrics, which create unpredictable trade-offs with each other and with predictive performance. It proposes a five-stage interdisciplinary framework: (1) identify legal requirements via legal analysis; (2) translate these into operationalizations and evaluation metrics/heuristics with combined legal and ML expertise; (3) form operationalization sets and train models; (4) evaluate through trade-off mapping across proxy metrics and performance; and (5) select a model and construct a documented legal justification, for example through a proportionality analysis. The framework is illustrated with a hypothetical anti-money laundering (AML) case study. The central claim is that the framework facilitates legal justification, provides insight into the impact of different operationalizations on trade-offs between proxy metrics and predictive performance, and enables organizations to select models that respect multiple legal obligations simultaneously.

Significance. If the framework works as claimed, it addresses a real gap at the intersection of law, software engineering, and machine learning: it provides a structured process for handling the indirect operationalization of legal obligations, the inherent uncertainty of legal interpretation, and the need for documented accountability. The paper makes a useful conceptual contribution by clearly distinguishing legal assessment from ethical assessment, by emphasizing that legal obligations require a holistic evaluation rather than a single-metric standard, and by stressing the role of interdisciplinary teams in metric selection and interpretation. It also correctly identifies that trade-offs among ethics-inspired proxy metrics are likely to emerge for legal obligations as well. However, the paper offers no empirical validation: the case study uses hypothetical numbers and assumed legal interpretations, and the framework includes no procedure for validating the legal import of the chosen proxy metrics. These limitations do not in themselves invalidate a methodological proposal, but they mean the operative claims should be read as a structured proposal rather than a demonstrated result.

major comments (3)
  1. [5.4, Table 2] The case study does not evaluate any trained model. The authors state, 'On the basis of insights from [67, 23, 68, 26, 71], Table 2 was generated,' meaning the trade-off values are invented for the illustration rather than derived from running the specified operationalization sets on data. Consequently, the Stage 5 selection of Set 3 and the associated legal justification are illustrative narratives, not conclusions produced by the framework applied to evidence. The paper's central claim that the framework 'provides insight into the impact of different operationalizations of legal obligations on trade-offs at the ML model level' is therefore not demonstrated; it is assumed by construction. The authors should either weaken this claim to a conceptual proposal or provide an empirical demonstration, at minimum a transparently labeled synthetic experiment with stated assumptions and sensitivity analysis.
  2. [4, Stage 2; 3.2] The framework selects proxy metrics in Stage 2 but provides no procedure for establishing that a chosen metric is a valid indicator of the corresponding legal obligation. The paper acknowledges in Section 3.2 that metrics are reductive and cannot prove compliance, but it does not operationalize that acknowledgment: there is no guidance on how to quantify the gap between a metric value and legal adequacy, how to choose among competing metrics that encode different legal interpretations, or how to detect when a metric gives false assurance. Because the Stage 4 trade-off maps and Stage 5 legal justification rely entirely on the selected proxies, the central claim that the framework 'facilitates legal justification' is weakened: a model can score well on all proxies while violating legal obligations, and the framework has no mechanism to surface that. Consider adding a validation sub-step, such as triangulation with doctrinal legal analysis, adversarial testing, or external audit.
  3. [5.5, Table 1, 5.2.4] There is a concrete internal inconsistency in the case study. Section 5.5 says 'the random forest model from Set 3 is chosen,' but Table 1 assigns Set 3 the AML Explainable Model operationalization (1), which Section 5.2.4 defines as logistic regression. This error renders the surrounding discussion of the relative explainability of random forests versus logistic regression incoherent, and it raises a question about whether the numeric trade-offs in Table 2 are associated with the correct models. The text and table should be reconciled.
minor comments (5)
  1. [3.2] Section 3.2 contains a duplicated word: 'ultimately focuses on the the ML model's alignment.' Please fix.
  2. [5.2.2] The phrase 'k-anonimity' is a typo for 'k-anonymity.' In addition, the 'low stopping threshold of -1.0e-07' is unexplained: the negative sign, units, and relationship to the Framework for Inhibiting Data Overcollection are not made clear.
  3. [Abstract, 1] The manuscript inconsistently uses 'Organizations' (Abstract) and 'organisations' (e.g., Section 1). Pick one spelling convention and apply it throughout.
  4. [Table 2] The column header 'CDD (Gender)' is not expanded in the table, and the relationship between the '% Data Used' values and the stated overcollection stopping threshold is not explained, so readers cannot see how the operationalizations translate into the reported numbers.
  5. [5.4] The statement that Table 2 was generated on the basis of insights from references [67, 23, 68, 26, 71] should specify which qualitative findings map to which numeric values; otherwise, the table is not reproducible or auditable.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the framework is a procedural workflow with no derived equations or fitted parameters, and its legal-justification output is explicitly acknowledged to be argumentative, not demonstrative.

full rationale

The paper proposes a five-stage workflow rather than a formal derivation, so there are no equations whose outputs reduce to their inputs and no fitted parameters that are later renamed as predictions. The central claim is that organizations can develop, evaluate, and select ML models with documented legal justification by choosing operationalizations and proxy metrics, mapping trade-offs, and then building a proportionality-based argument. The only point that might look self-referential is that the Stage 5 legal justification is built from the same proxy metrics and operationalizations that the team itself selected in Stage 2, and the chosen model is justified by its scores on those metrics. However, the paper repeatedly disclaims that proxy metrics prove legal compliance, for example: 'it cannot show no legal obligations have been violated' and 'as metrics are aggregates, they inherently obscure individual contexts.' The claimed output is therefore an argument supporting legal alignment, not a proof of compliance, and the paper does not equate a favorable metric value with satisfaction of the legal obligation. The authors' self-citations are incidental and none carries a load-bearing uniqueness or derivation claim. Table 2 is explicitly said to have been generated on the basis of literature insights rather than real model runs, so the case study is presented as illustrative, not as an empirical prediction. In sum, the contribution is organizational and interdisciplinary; no circular reduction is present.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on translation and transferability assumptions about legal obligations and proxy metrics. No new entities are introduced. The case study parameters are illustrative, not fitted.

free parameters (3)
  • k-anonymity parameter k = 7 (case study)
    Chosen a priori in the AML case study (Sections 5.2.2 and 5.2.3) as suitable for data protection without empirical justification; illustrative, not tuned to data.
  • Data overcollection stopping threshold = -1.0e-07
    Selected in the case study (Section 5.2.2) as a low threshold, with the rationale that performance is crucial for AML obligations; illustrative.
  • Cost-sensitive positive class weight = not specified
    The AML risk coverage operationalization assigns higher weights to positive instances (Section 5.2.5), but no value is given; implementation-dependent.
assumptions (4)
  • domain assumption Legal obligations can be converted into operationalizations and proxy metrics that are meaningful for legal evaluation.
    The entire framework relies on this translation being possible (Sections 3 and 4). The authors note metrics are reductive and cannot prove compliance, but the framework still depends on them being informative.
  • domain assumption Trade-offs observed among fairness, interpretability, and privacy metrics transfer to legal-proxy metrics.
    Section 3 argues similar trade-offs will emerge for legal obligations based on ethics and ML literature [23,67,68,26,71]. This is an extrapolation from empirical findings in other domains.
  • domain assumption The legal interpretations adopted in the case study are defensible estimates of what authorities would decide.
    Sections 5.1 and 5.2 make contestable legal choices (e.g., direct discrimination on gender, GDPR Article 22 inapplicability due to human intervention, models as personal data). The framework depends on these stances being accepted as legally grounded.
  • ad hoc to paper The illustrative numbers in Table 2 are representative of real trade-offs.
    Table 2 is generated from literature insights, not experiments; it is used to demonstrate the framework's mechanics, not to support empirical claims about real-world performance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Engineering the Law-Machine Learning Translation Problem: Developing Legally Aligned Models." pith.science (2026). https://pith.science/paper/XTRPKQSH

@misc{pith2026250416969,
  author       = {Pith},
  title        = {Pith review of: Engineering the Law-Machine Learning Translation Problem: Developing Legally Aligned Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XTRPKQSH}},
  note         = {Machine review of arXiv:2504.16969}
}
read the original abstract

Organizations developing machine learning-based (ML) technologies face the complex challenge of achieving high predictive performance while respecting the law. This intersection between ML and the law creates new complexities. As ML model behavior is inferred from training data, legal obligations cannot be operationalized in source code directly. Rather, legal obligations require "indirect" operationalization. However, choosing context-appropriate operationalizations presents two compounding challenges: (1) laws often permit multiple valid operationalizations for a given legal obligation-each with varying degrees of legal adequacy; and, (2) each operationalization creates unpredictable trade-offs among the different legal obligations and with predictive performance. Evaluating these trade-offs requires metrics (or heuristics), which are in turn difficult to validate against legal obligations. Current methodologies fail to fully address these interwoven challenges as they either focus on legal compliance for traditional software or on ML model development without adequately considering legal complexities. In response, we introduce a five-stage interdisciplinary framework that integrates legal and ML-technical analysis during ML model development. This framework facilitates designing ML models in a legally aligned way and identifying high-performing models that are legally justifiable. Legal reasoning guides choices for operationalizations and evaluation metrics, while ML experts ensure technical feasibility, performance optimization and an accurate interpretation of metric values. This framework bridges the gap between more conceptual analysis of law and ML models' need for deterministic specifications. We illustrate its application using a case study in the context of anti-money laundering.

Figures

Figures reproduced from arXiv: 2504.16969 by the authors.

Figure 1
Figure 1. 5-Stage Framework to Design, Evaluate and Select ML Models under Legal Obligations [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Decision-centric fairness: Evaluation and optimization for resource allocation problems

    cs.LG 2025-04 conditional novelty 4.0 of 10

    Restricting demographic parity enforcement to the decision-making region of a score distribution yields better fairness-performance trade-offs than enforcing it globally.

Reference graph

Works this paper leans on

122 extracted references · 69 canonical work pages · cited by 1 Pith paper

  1. [1]

    Battaglini, L

    M. Battaglini, L. Guiso, C. Lacava, D. L. Miller, E. Patacchini, Refining public policies with machine learning: The case of tax auditing, Journal of Econometrics (2024) 105847

    2024

  2. [2]

    De Roux, B

    D. De Roux, B. Perez, A. Moreno, M. d. P. Villamil, C. Figueroa, Tax fraud detection for under-reporting declarations using an unsupervised machine learning approach, in: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2018, pp. 215–222

    2018

  3. [3]

    Z. Chen, L. D. Van Khoa, E. N. Teoh, A. Nazir, E. K. Karuppiah, K. S. Lam, Machine learning techniques for anti-money laundering (aml) solutions in suspicious transaction detection: a review, Knowledge and Information Systems 57 (2018) 245–285

    2018

  4. [4]

    Rahman, V

    M. Rahman, V . Kumar, Machine learning based customer churn prediction in banking, in: 2020 4th international conference on electronics, communication and aerospace technology (ICECA), IEEE, 2020, pp. 1196–1201

    2020

  5. [5]

    De Caigny, K

    A. De Caigny, K. W. De Bock, S. Verboven, Hybrid black-box classification for customer churn prediction with segmented interpretability analysis, Decision Support Systems 181 (2024) 114217

    2024

  6. [6]

    Verboven, N

    S. Verboven, N. Martin, Combining the clinical and operational perspectives in heterogeneous treatment e ffect inference in healthcare processes, in: International Conference on Process Mining, Springer, 2021, pp. 327–339

    2021

  7. [7]

    Malgieri, F

    G. Malgieri, F. Pasquale, Licensing high-risk artificial intelligence: Toward ex ante justification for a disruptive technology, Computer Law & Security Review 52 (2024) 105899

    2024

  8. [8]

    Asgeirsson, The nature and value of vagueness in the law, Bloomsbury Publishing, 2020

    H. Asgeirsson, The nature and value of vagueness in the law, Bloomsbury Publishing, 2020

    2020

Show all 122 references
  1. [9]

    Hildebrandt, The adaptive nature of text-driven law, Journal of Cross-disciplinary Research in Computational Law 1 (2021)

    M. Hildebrandt, The adaptive nature of text-driven law, Journal of Cross-disciplinary Research in Computational Law 1 (2021)

    2021

  2. [10]

    Weerts, R

    H. Weerts, R. Xenidis, F. Tarissan, H. P. Olsen, M. Pechenizkiy, Algorithmic unfairness through the lens of eu non-discrimination law: Or why the law is not a decision tree, in: Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, 2023, pp. 805–816

    2023

  3. [11]

    Dari-Mattiacci, B

    G. Dari-Mattiacci, B. De ffains, Uncertainty of law and the legal process, Journal of Institutional and Theoretical Economics (JITE)/Zeitschrift f¨ur die gesamte Staatswissenschaft (2007) 627–656

    2007

  4. [12]

    Siena, J

    A. Siena, J. Mylopoulos, A. Perini, A. Susi, Designing law-compliant software requirements, in: Conceptual Modeling-ER 2009: 28th International Conference on Conceptual Modeling, Gramado, Brazil, November 9-12, 2009. Proceedings 28, Springer, 2009, pp. 472–486

    2009

  5. [13]

    P. N. Otto, A. I. Ant ´on, Addressing legal requirements in requirements engineering, in: 15th IEEE international requirements engineering conference (RE 2007), IEEE, 2007, pp. 5–14

    2007

  6. [14]

    J. E. Calfee, R. Craswell, Some e ffects of uncertainty on compliance with legal standards, Va. L. Rev. 70 (1984) 965

    1984

  7. [15]

    Hardt, E

    M. Hardt, E. Price, N. Srebro, Equality of opportunity in supervised learning, Advances in neural information processing systems 29 (2016)

    2016

  8. [16]

    Shanmugam, F

    D. Shanmugam, F. Diaz, S. Shabanian, M. Finck, A. Biega, Learning to limit data collection via scaling laws: A computational interpretation for the legal principle of data minimization, in: Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, 2...

    2022

  9. [17]

    Goldsteen, G

    A. Goldsteen, G. Ezov, R. Shmelkin, M. Mo ffie, A. Farkash, Data minimization for gdpr compliance in machine learning models, AI and Ethics 2 (2022) 477–491

    2022

  10. [18]

    Goldsteen, O

    A. Goldsteen, O. Saadi, R. Shmelkin, S. Shachor, N. Razinkov, Ai privacy toolkit, SoftwareX 22 (2023) 101352

    2023

  11. [19]

    Rudin, Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead, Nature machine intelligence 1 (2019) 206–215

    C. Rudin, Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead, Nature machine intelligence 1 (2019) 206–215

    2019

  12. [20]

    Fresz, E

    B. Fresz, E. Dubovitskaya, D. Brajovic, M. F. Huber, C. Horz, How should ai decisions be explained? requirements for explanations from the perspective of european law, in: Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, volume 7, 2024, pp. 438–450

    2024

  13. [21]

    Wachter, B

    S. Wachter, B. Mittelstadt, C. Russell, Why fairness cannot be automated: Bridging the gap between eu non-discrimination law and ai, Computer Law & Security Review 41 (2021) 105567. URL: https://www.sciencedirect.com/science/article/pii/ S0267364921000406. doi:https://doi.org/...

    2021

  14. [22]

    P. K. Bandara, H. D. Bandara, S. Fernando, Evaluation of re-identification risks in data anonymization techniques based on population uniqueness, in: 2020 5th International Conference on Information Technology Research (ICITR), IEEE, 2020, pp. 1–5

    2020

  15. [23]

    Agarwal, Trade-o ffs between fairness and privacy in machine learning, in: IJCAI 2021 Workshop on AI for Social Good, 2021

    S. Agarwal, Trade-o ffs between fairness and privacy in machine learning, in: IJCAI 2021 Workshop on AI for Social Good, 2021

    2021

  16. [24]

    X. Gu, Z. Tianqing, J. Li, T. Zhang, W. Ren, K.-K. R. Choo, Privacy, accuracy, and model fairness trade-o ffs in federated learning, Computers & Security 122 (2022) 102907

    2022

  17. [25]

    K. Sun, X. Zhang, X. Lin, G. Li, J. Wang, J. Li, Toward the tradeo ffs between privacy, fairness and utility in federated learning, 2023. arXiv:2311.18190. 13

    2023 arXiv

  18. [26]

    Ferry, U

    J. Ferry, U. A ¨ıvodji, S. Gambs, M.-J. Huguet, M. Siala, Sok: Taming the triangle–on the interplays between fairness, interpretability and privacy in machine learning, arXiv preprint arXiv:2312.16191 (2023)

    2023 arXiv

  19. [27]

    Pawelczyk, T

    M. Pawelczyk, T. Leemann, A. Biega, G. Kasneci, On the trade-o ff between actionable explanations and the right to be forgotten, 2023. arXiv:2208.14137

    2023 arXiv

  20. [28]

    Xhemajli, The role of ethics and morality in law: Similarities and di fferences, Ohio NUL Rev

    H. Xhemajli, The role of ethics and morality in law: Similarities and di fferences, Ohio NUL Rev. 48 (2021) 81

    2021

  21. [29]

    Gardner, 35 ethics and law, The Routledge Companion to Ethics (2010)

    J. Gardner, 35 ethics and law, The Routledge Companion to Ethics (2010)

    2010

  22. [30]

    C. A. Rentmeester, What’s legal? what’s moral? what’s the di fference? a guide for teaching residents, The American Journal of Bioethics 6 (2006) 31–33

    2006

  23. [31]

    Lewkowicz, R

    G. Lewkowicz, R. Sarf, Taking technical standardization of fundamental rights seriously for trustworthy artificial intelligence, La Revue des Juristes de Sciences Po (2024) 42–46. URL: https://centreperelman.be//content/uploads/2024/03/GL-RS-AI-Technical- standards-and-fundame...

    2024

  24. [32]

    A. D. Selbst, D. Boyd, S. A. Friedler, S. Venkatasubramanian, J. Vertesi, Fairness and abstraction in sociotechnical systems, in: Proceedings of the conference on fairness, accountability, and transparency, 2019, pp. 59–68

    2019

  25. [33]

    Blackman, Ethical machines: Your concise guide to totally unbiased, transparent, and respectful AI, Harvard Business Press, 2022

    R. Blackman, Ethical machines: Your concise guide to totally unbiased, transparent, and respectful AI, Harvard Business Press, 2022

    2022

  26. [34]

    L. K. Koumeri, M. Legast, Y . Yousefi, K. Vanhoof, A. Legay, C. Schommer, Compatibility of fairness metrics with eu non-discrimination laws: Demographic parity & conditional demographic disparity, arXiv preprint arXiv:2306.08394 (2023)

    2023 arXiv

  27. [35]

    Kerrigan, Artificial intelligence: Law and regulation, Edward Elgar Publishing, 2022

    C. Kerrigan, Artificial intelligence: Law and regulation, Edward Elgar Publishing, 2022

    2022

  28. [36]

    Hadjiemmanuil, A heavily regulated industry, eucrim (2015)

    C. Hadjiemmanuil, A heavily regulated industry, eucrim (2015)

    2015

  29. [37]

    Shen, Ai regulation in health care: How washington state can conquer the new territory of ai regulation, Seattle J

    N. Shen, Ai regulation in health care: How washington state can conquer the new territory of ai regulation, Seattle J. Tech. Env’t & Innovation L. 13 (2023) 1

    2023

  30. [38]

    Hassani, M

    S. Hassani, M. Sabetzadeh, D. Amyot, J. Liao, Rethinking legal compliance automation: Opportunities with large language models, arXiv preprint arXiv:2404.14356 (2024)

    2024 arXiv

  31. [39]

    Sleimi, M

    A. Sleimi, M. Ceci, N. Sannier, M. Sabetzadeh, L. Briand, J. Dann, A query system for extracting requirements-related information from legal texts, in: 2019 IEEE 27th international requirements engineering conference (RE), IEEE, 2019, pp. 319–329

    2019

  32. [40]

    Akhigbe, D

    O. Akhigbe, D. Amyot, G. Richards, A systematic literature mapping of goal and non-goal modelling methods for legal and regulatory compliance, Requirements Engineering 24 (2019) 459–481

    2019

  33. [41]

    Abualhaija, M

    S. Abualhaija, M. Ceci, L. Briand, Legal requirements analysis, arXiv preprint arXiv:2311.13871 (2023)

    2023 arXiv

  34. [42]

    Faßbender, Domain-and Quality-aware Requirements Engineering for Law-compliant Systems, Ph.D

    S. Faßbender, Domain-and Quality-aware Requirements Engineering for Law-compliant Systems, Ph.D. thesis, Dissertation, Duisburg, Essen, Universit¨at Duisburg-Essen, 2017, 2017

    2017

  35. [43]

    Hoess, N

    A. Hoess, N. Pocher, T. H. Roth, G. Fridgen, Towards a design science research process for legal compliance by design, in: PACIS 2024 Proceedings, 2024, p. 1609

    2024

  36. [44]

    Gjorgjevikj, K

    A. Gjorgjevikj, K. Mishev, L. Antovski, D. Trajanov, Requirements engineering in machine learning projects, IEEE Access (2023)

    2023

  37. [45]

    Zowghi, M

    D. Zowghi, M. Bano, What’s missing in requirements engineering for responsible ai?, IEEE Software 40 (2023) 11–15

    2023

  38. [46]

    E. K. Budiardjo, et al., Roadmap analysis of artificial intelligence engineering method., Revue d’Intelligence Artificielle 38 (2024)

    2024

  39. [47]

    Ahmad, M

    K. Ahmad, M. Abdelrazek, C. Arora, M. Bano, J. Grundy, Requirements engineering for artificial intelligence systems: A systematic mapping study, Information and Software Technology 158 (2023) 107176

    2023

  40. [48]

    Mart ´ınez-Fern´andez, J

    S. Mart ´ınez-Fern´andez, J. Bogner, X. Franch, M. Oriol, J. Siebert, A. Trendowicz, A. M. V ollmer, S. Wagner, Software engineering for ai-based systems: a survey, ACM Transactions on Software Engineering and Methodology (TOSEM) 31 (2022) 1–59

    2022

  41. [49]

    De Martino, F

    V . De Martino, F. Palomba, Classification, challenges, and automated approaches to handle non-functional requirements in ml-enabled systems: A systematic literature review, arXiv preprint arXiv:2311.17483 (2023)

    2023 arXiv

  42. [50]

    Boella, L

    G. Boella, L. Humphreys, R. Muthuri, P. Rossi, L. van der Torre, A critical analysis of legal requirements engineering from the perspective of legal practice, in: 2014 IEEE 7th International Workshop on Requirements Engineering and Law (RELAW), IEEE, 2014, pp. 14–21

    2014

  43. [51]

    doi: 10.1109/IEEESTD.1990.101064

    IEEE, Ieee standard glossary of software engineering terminology, 1990. doi: 10.1109/IEEESTD.1990.101064

    1990

  44. [52]

    Van Vliet, H

    H. Van Vliet, H. Van Vliet, J. Van Vliet, Software engineering: principles and practice, volume 13, John Wiley & Sons Hoboken, NJ, 2008

    2008

  45. [53]

    Belani, M

    H. Belani, M. Vukovic, ˇZ. Car, Requirements engineering challenges in building ai-based complex systems, in: 2019 IEEE 27th International Requirements Engineering Conference Workshops (REW), IEEE, 2019, pp. 252–255

    2019

  46. [54]

    Hildebrandt, A

    M. Hildebrandt, A. De Bois, Law for computer scientists, in: ECCAI Advanced Course on Artificial Intelligence, Springer, 2021, pp. 261–274

    2021

  47. [55]

    Merriam-Webster, Merriam-webster.com dictionary, https://www.merriam-webster.com/dictionary/law, 2025

    2025

  48. [56]

    Ghanavati, D

    S. Ghanavati, D. Amyot, A. Siena, A. Susi, A. Perini, Making business processes law compliant, in: First workshop on law compliancy issues in organisational systems and strategies (iComply’10). Retrieved, volume 5, 2016

    2016

  49. [57]

    J. P. Castellanos Ardila, B. Gallina, F. Ul Muram, Compliance checking of software processes: A systematic literature review, Journal of Software: Evolution and Process 34 (2022) e2440

    2022

  50. [58]

    Ahmad, M

    K. Ahmad, M. Abdelrazek, C. Arora, J. Grundy, M. Bano, Requirements elicitation and modelling of artificial intelligence systems: An empirical study, arXiv preprint arXiv:2302.06034 (2023)

    2023 arXiv

  51. [59]

    Ahmad, M

    K. Ahmad, M. Abdelrazek, C. Arora, A. A. Baniya, M. Bano, J. Grundy, Requirements engineering framework for human-centered artificial intelligence software systems, Applied Soft Computing 143 (2023) 110455

    2023

  52. [60]

    Ahmad, C

    K. Ahmad, C. Arora, M. Abdelrazek, J. Grundy, R. Vasa, Requirements elicitation in the age of ai: A tool’s multi-system journey, in: International Conference on Evaluation of Novel Approaches to Software Engineering, Springer, 2023, pp. 67–90

    2023

  53. [61]

    M. Bano, D. Zowghi, V . Gervasi, R. Shams, Ai for all: Operationalising diversity and inclusion requirements for ai systems, arXiv preprint arXiv:2311.14695 (2023)

    2023 arXiv

  54. [62]

    T. Sporsem, Discovering explainability requirements in ml-based software, in: Proceedings of the 2024 IEEE /ACM 46th International Conference on Software Engineering: Companion Proceedings, 2024, pp. 170–172

    2024

  55. [63]

    Balasubramaniam, M

    N. Balasubramaniam, M. Kauppinen, H.-L. Truong, S. Kujala, Candidate solutions for defining explainability requirements of ai systems, in: International Working Conference on Requirements Engineering: Foundation for Software Quality, Springer, 2024, pp. 129–146. 14

    2024

  56. [64]

    Damirchi, A

    R. Damirchi, A. Amini, Non-functional requirement extracting methods for ai-based systems: A survey, in: 2023 13th International Conference on Computer and Knowledge Engineering (ICCKE), IEEE, 2023, pp. 535–539

    2023

  57. [65]

    M. C. Martins, T. Novo, A requirements engineering process for machine learning innovation projects, in: Proceedings of the 27th Workshop on Requirements Engineering (WER24), 2024

    2024

  58. [66]

    Sadovski, I

    E. Sadovski, I. Aviv, I. Hadar, Towards a comprehensive ontology for requirements engineering for ai-powered systems, in: International Working Conference on Requirements Engineering: Foundation for Software Quality, Springer, 2024, pp. 219–230

    2024

  59. [67]

    Agarwal, Trade-o ffs between fairness and interpretability in machine learning, in: IJCAI 2021 Workshop on AI for Social Good, 2021, pp

    S. Agarwal, Trade-o ffs between fairness and interpretability in machine learning, in: IJCAI 2021 Workshop on AI for Social Good, 2021, pp. 1–6

    2021

  60. [68]

    Jabbari, H.-C

    S. Jabbari, H.-C. Ou, H. Lakkaraju, M. Tambe, An empirical study of the trade-o ffs between interpretability and fairness, in: ICML Workshop on Human Interpretability in Machine Learning, International Conference on Machine Learning (ICML), 2020

    2020

  61. [69]

    N. Jo, S. Aghaei, A. G ´omez, P. Vayanos, Learning optimal fair classification trees: Trade-o ffs between interpretability, fairness, and accuracy, arXiv preprint arXiv:2201.09932 (2022)

    2022 arXiv

  62. [70]

    Pannekoek, G

    M. Pannekoek, G. Spigler, Investigating trade-o ffs in utility, fairness and di fferential privacy in neural networks, arXiv preprint arXiv:2102.05975 (2021)

    2021 arXiv

  63. [71]

    Gittens, B

    A. Gittens, B. Yener, M. Yung, An adversarial perspective on accuracy, robustness, fairness, and privacy: Multilateral-tradeoffs in trustwor- thy ml, IEEE Access 10 (2022) 120850–120865

    2022

  64. [72]

    K. T. Rodolfa, H. Lamba, R. Ghani, Empirical observation of negligible fairness–accuracy trade-o ffs in machine learning for public policy, Nature Machine Intelligence 3 (2021) 896–904

    2021

  65. [73]

    Finck, A

    M. Finck, A. J. Biega, Reviving purpose limitation and data minimisation in data-driven systems, Technology and Regulation (2021) V ol. 2021 (2021). URL: https://techreg.org/article/view/10986. doi:10.26116/TECHREG.2021.004

    2021 doi

  66. [74]

    Pessach, E

    D. Pessach, E. Shmueli, A review on fairness in machine learning, ACM Computing Surveys (CSUR) 55 (2022) 1–44

    2022

  67. [75]

    K. J. Vandevelde, Thinking like a lawyer: An introduction to legal reasoning, Routledge, 2018

    2018

  68. [76]

    Bueno Mom ˇcilovi´c, D

    T. Bueno Mom ˇcilovi´c, D. Balta, Challenges of assuring compliance of information systems in finance, in: International Conference on Software Quality, Springer, 2024, pp. 135–152

    2024

  69. [77]

    M. E. Kaminski, G. Malgieri, Algorithmic impact assessments under the GDPR: producing multi-layered explanations, HeinOnline, 2020

    2020

  70. [78]

    right to explanation

    B. Goodman, S. Flaxman, European union regulations on algorithmic decision-making and a “right to explanation”, AI magazine 38 (2017) 50–57

    2017

  71. [79]

    meaningful information

    A. Selbst, J. Powles, “meaningful information” and the right to explanation, in: conference on fairness, accountability and transparency, PMLR, 2018, pp. 48–48

    2018

  72. [80]

    Malgieri, G

    G. Malgieri, G. Comand ´e, Why a right to legibility of automated decision-making exists in the general data protection regulation, Interna- tional Data Privacy Law 7 (2017) 243–265

    2017

  73. [81]

    right to explanation

    E. Bayamlıo ˘glu, The right to contest automated decisions under the general data protection regulation: Beyond the so-called “right to explanation”, Regulation & Governance 16 (2022) 1058–1078

    2022

  74. [82]

    Wachter, B

    S. Wachter, B. Mittelstadt, L. Floridi, Why a right to explanation of automated decision-making does not exist in the general data protection regulation, International data privacy law 7 (2017) 76–99

    2017

  75. [83]

    Custers, New digital rights: Imagining additional fundamental rights for the digital era, Computer Law & Security Review 44 (2022) 105636

    B. Custers, New digital rights: Imagining additional fundamental rights for the digital era, Computer Law & Security Review 44 (2022) 105636

    2022

  76. [84]

    Tridimas, Wreaking the wrongs: Balancing rights and the public interest the eu way, Colum

    T. Tridimas, Wreaking the wrongs: Balancing rights and the public interest the eu way, Colum. J. Eur. L. 29 (2023) 185

    2023

  77. [85]

    R. Gellert, We have always managed risks in data protection law: understanding the similarities and di fferences between the rights-based and the risk-based approaches to data protection, Eur. Data Prot. L. Rev. 2 (2016) 481

    2016

  78. [86]

    T. C. Lin, Artificial intelligence, finance, and the law, Fordham L. Rev. 88 (2019) 531

    2019

  79. [87]

    R. I. T. Jensen, A. Iosifidis, Fighting money laundering with statistics and machine learning, IEEE Access 11 (2023) 8889–8903

    2023

  80. [88]

    European Union, Regulation (eu) 2024 /1689 laying down harmonised rules on artificial intelligence (i.e., the artificial intelligence act, https://www.europarl.europa.eu/doceo/document/TA-9-2024-0138_EN.pdf , 2024

    2024

  81. [89]

    Parente, The ai act and its impacts on the european financial sector, The EUROFI Magazine (2024)

    F. Parente, The ai act and its impacts on the european financial sector, The EUROFI Magazine (2024). Ghent 2024, DIGITALISATION AND TECHNOLOGY , VIEWS, 128

    2024

  82. [90]

    Xenidis, L

    R. Xenidis, L. Senden, Eu non-discrimination law in the era of artificial intelligence: Mapping the challenges of algorithmic discrimination, Rapha¨ele Xenidis and Linda Senden,‘EU non-discrimination law in the era of artificial intelligence: Mapping the challenges of algorith...

    2019

  83. [91]

    URL: https://fra.europa.eu/ en/publication/2018/handbook-european-non-discrimination-law-2018-edition , accessed: date-of-access

    European Union Agency for Fundamental Rights, European Court of Human Rights, Council of Europe, Handbook on European non- discrimination law, 2018 ed., European Union Agency for Fundamental Rights, Vienna, Austria, 2018. URL: https://fra.europa.eu/ en/publication/2018/handboo...

    2018

  84. [92]

    M. De Mol, The novel approach of the cjeu on the horizontal direct effect of the eu principle of non-discrimination:(unbridled) expansionism of eu law?, Maastricht journal of European and Comparative law 18 (2011) 109–135

    2011

  85. [93]

    Muir, The horizontal e ffects of charter rights given expression to in eu legislation, from mangold to bauer, Review of European Administrative Law 12 (2019) 185–215

    E. Muir, The horizontal e ffects of charter rights given expression to in eu legislation, from mangold to bauer, Review of European Administrative Law 12 (2019) 185–215

    2019

  86. [94]

    Bertrand, W

    A. Bertrand, W. Maxwell, X. Vamparys, Do ai-based anti-money laundering (aml) systems violate european fundamental rights?, Interna- tional data privacy law 11 (2021) 276–293

    2021

  87. [95]

    L. L. Sciurba, Michele, The Incompatibility of Global Anti-Money Laundering Regimes with Human and Civil Rights, 1 ed., Nomos, Baden-Baden, 2019

    2019

  88. [96]

    Article 5.1(c)

    European Union, General Data Protection Regulation, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX% 3A32016R0679, 2016. Article 5.1(c)

    2016

  89. [97]

    Binns, M

    R. Binns, M. Veale, Is that your final decision? multi-stage profiling, selective e ffects, and article 22 of the gdpr, International Data Privacy Law 11 (2021) 319–332. 15

    2021

  90. [98]

    C. of Justice of the European Union, Judgment of the court of justice of the european union (first chamber) of 7 december 2023: Oq v land hessen, schufa holding ag (case c-634 /21), 2023. URL: https://curia.europa.eu/juris/document/document.jsf?mode= DOC&pageIndex=0&docid=2804...

    2023

  91. [99]

    K. Shrishak, Ai-complex algorithms and e ffective data protection supervision - e ffective implementation of data subjects’ rights, https: //www.edpb.europa.eu/system/files/2025-01/d2-ai-effective-implementation-of-data-subjects-rights_en.pdf ,

    2025

  92. [100]

    Veale, R

    M. Veale, R. Binns, L. Edwards, Algorithms that remember: model inversion attacks and data protection law, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 376 (2018) 20180083

    2018

  93. [101]

    B. A. Juliussen, J. P. Rui, D. Johansen, Algorithms that forget: Machine unlearning and the right to erasure, Computer Law & Security Review 51 (2023) 105885

    2023

  94. [102]

    Leiser, F

    M. Leiser, F. Dechesne, Governing machine-learning models: challenging the personal data presumption, International data privacy law 10 (2020) 187–200

    2020

  95. [103]

    C. King, C. Walker, J. Gurul ´e, The Palgrave handbook of criminal and terrorism financing law, Springer, 2018

    2018

  96. [104]

    E. Commission, Anti-money laundering and countering the financing of terrorism at eu level, https://finance.ec.europa.eu/ financial-crime/anti-money-laundering-and-countering-financing-terrorism-eu-level_en , 2024. Accessed on 29 May 2024

    2024

  97. [105]

    Kamiran, A

    F. Kamiran, A. Karim, X. Zhang, Decision theory for discrimination-aware classification, in: 2012 IEEE 12th international conference on data mining, IEEE, 2012, pp. 924–929

    2012

  98. [106]

    Kenfack, Exploring the landscape of ai ethics, https://www.kaggle.com/code/patrikkenfack/exploring-the-landscape- of-ai-ethics/notebook , 2023

    P. Kenfack, Exploring the landscape of ai ethics, https://www.kaggle.com/code/patrikkenfack/exploring-the-landscape- of-ai-ethics/notebook , 2023. Accessed on 21 May 2024

    2023

  99. [107]

    Wachter, B

    S. Wachter, B. Mittelstadt, C. Russell, Bias preservation in machine learning: the legality of fairness metrics under eu non-discrimination law, W. Va. L. Rev. 123 (2020) 735

    2020

  100. [108]

    C. Quelle, Enhancing compliance under the general data protection regulation: the risky upshot of the accountability-and risk-based approach, European Journal of Risk Regulation 9 (2018) 502–526

    2018

  101. [109]

    Shokri, M

    R. Shokri, M. Stronati, C. Song, V . Shmatikov, Membership inference attacks against machine learning models, in: 2017 IEEE symposium on security and privacy (SP), IEEE, 2017, pp. 3–18

    2017

  102. [110]

    Jegorova, C

    M. Jegorova, C. Kaul, C. Mayor, A. Q. O’Neil, A. Weir, R. Murray-Smith, S. A. Tsaftaris, Survey: Leakage and privacy at inference time, IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)

    2022

  103. [111]

    S. Yeom, I. Giacomelli, M. Fredrikson, S. Jha, Privacy risk in machine learning: Analyzing the connection to overfitting, in: 2018 IEEE 31st computer security foundations symposium (CSF), IEEE, 2018, pp. 268–282

    2018

  104. [112]

    C. A. Choquette-Choo, F. Tramer, N. Carlini, N. Papernot, Label-only membership inference attacks, in: M. Meila, T. Zhang (Eds.), Proceedings of the 38th International Conference on Machine Learning, volume 139 ofProceedings of Machine Learning Research, PMLR, 2021, pp. 1964–1...

    2021

  105. [113]

    S. V . Dibbo, D. L. Chung, S. Mehnaz, Model inversion attack with least information and an in-depth analysis of its disparate vulnerability, in: 2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML), IEEE, 2023, pp. 119–135

    2023

  106. [114]

    Kobayashi, S

    K. Kobayashi, S. B. Alam, Explainable, interpretable, and trustworthy ai for an intelligent digital twin: A case study on remaining useful life, Engineering Applications of Artificial Intelligence 129 (2024) 107620

    2024

  107. [115]

    L.-V . Herm, K. Heinrich, J. Wanner, C. Janiesch, Stop ordering machine learning algorithms by their explainability! a user-centered investigation of performance and explainability, International Journal of Information Management 69 (2023) 102538

    2023

  108. [116]

    Fern ´andez, S

    A. Fern ´andez, S. Garc´ıa, M. Galar, R. C. Prati, B. Krawczyk, F. Herrera, A. Fern´andez, S. Garc´ıa, M. Galar, R. C. Prati, et al., Cost-sensitive learning, Learning from imbalanced data sets (2018) 63–78

    2018

  109. [117]

    Hoang, K

    D. Hoang, K. Wiegratz, Machine learning methods in finance: Recent applications and prospects, European Financial Management 29 (2023) 1657–1701

    2023

  110. [118]

    Bockel-Rickermann, S

    C. Bockel-Rickermann, S. Verboven, T. Verdonck, W. Verbeke, A causal perspective on loan pricing: Investigating the impacts of selection bias on identifying bid-response functions, arXiv preprint arXiv:2309.03730 (2023)

    2023 arXiv

  111. [119]

    Giovanola, S

    B. Giovanola, S. Tiribelli, Beyond bias and discrimination: redefining the ai ethics principle of fairness in healthcare machine-learning algorithms, AI & society 38 (2023) 549–563

    2023

  112. [120]

    Gilbert, The eu passes the ai act and its implications for digital medicine are unclear, npj Digital Medicine 7 (2024) 135

    S. Gilbert, The eu passes the ai act and its implications for digital medicine are unclear, npj Digital Medicine 7 (2024) 135

    2024

  113. [121]

    Rasheed, A

    K. Rasheed, A. Qayyum, M. Ghaly, A. Al-Fuqaha, A. Razi, J. Qadir, Explainable, trustworthy, and ethical machine learning for healthcare: A survey, Computers in Biology and Medicine 149 (2022) 106043. 16

    2022

  114. [2025]

    Accessed on 6 March 2025

    2025

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.