The AI risk repository: A meta-review, database, and taxonomy of risks from artificial intelligence

Alexander K. Saeri; Emily A. C. Grundy; James Dao; Jess Graham; Michael Noetel; Neil Thompson; Peter Slattery; Risto Uuk; Soroush Pour; Stephen Casper

arxiv: 2408.12622 · v3 · submitted 2024-08-14 · 💻 cs.AI · cs.CR· cs.ET· cs.LG· cs.SY· eess.SY

The AI risk repository: A meta-review, database, and taxonomy of risks from artificial intelligence

Peter Slattery , Alexander K. Saeri , Emily A. C. Grundy , Jess Graham , Michael Noetel , Risto Uuk , James Dao , Soroush Pour

show 2 more authors

Stephen Casper Neil Thompson

This is my paper

Pith reviewed 2026-05-23 21:54 UTC · model grok-4.3

classification 💻 cs.AI cs.CRcs.ETcs.LGcs.SYeess.SY

keywords AI risksrisk taxonomymeta-reviewAI safetyrisk databaseclassification systemhuman-AI risk sources

0 comments

The pith

Unified AI risk catalog from 74 frameworks shows human decisions cause 38% of risks

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

The paper collects 1,725 distinct risks from every major AI risk framework published to date, which totals 74 frameworks. It organizes these risks into a single repository using two new classification systems. One system attributes risks to human decisions or to the AI systems themselves and finds humans responsible for 38% while the systems account for 42%. This addresses inconsistent terminology across studies, such as varying definitions of privacy or multiple terms for the same misalignment issue. The resulting database serves as a shared reference for developers doing risk assessments, policymakers drafting rules, and auditors reviewing systems.

Core claim

By systematically reviewing 74 AI risk frameworks containing 1,725 distinct risks, the authors built a unified repository with two classification systems. These systems organize the risks under shared categories and reveal that human decisions cause nearly as many AI risks (38%) as the AI systems themselves (42%). The repository supplies a common reference point that reduces the need to map between separate taxonomies and supports more complete coverage of risks.

What carries the argument

Dual classification systems that map 1,725 risks from 74 frameworks into one database, with one system separating risks caused by human decisions from those caused by AI systems.

If this is right

Risk assessments can reference one comprehensive source instead of consulting multiple separate frameworks.
Regulations can be drafted against a shared list of risks without gaps from differing terminology.
Audits of AI systems can evaluate against the full mapped set of risks in a consistent way.
Research findings can be compared across studies without first translating between different taxonomies.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The repository could function as a living document that incorporates new frameworks as they appear.
The balance between human and AI causes points to a need for stronger controls on human choices during AI design and deployment.
Widespread adoption might encourage standardized risk reporting requirements across the AI industry.

Load-bearing premise

The 74 frameworks represent all major AI risk frameworks published to date and the mapping of the 1,725 risks into the new classifications preserves original meanings without significant selection or interpretation bias.

What would settle it

Discovery of a major AI risk framework published before the cutoff that was omitted from the 74, or an audit showing that a substantial portion of risks had their meanings altered during the mapping process.

read the original abstract

Artificial intelligence (AI) is reshaping society, from video generation to medical diagnosis, coding agents to autonomous vehicles. Yet researchers, policymakers, and technology companies lack shared terminology for discussing AI risks. Consider "privacy": one framework uses this term to describe a model's ability to leak sensitive training data, while another uses it to mean freedom from government surveillance. Conversely, researchers have introduced "Goodhart's law," "specification gaming," "reward hacking," and "mesa-optimization" to describe the same phenomenon of AI systems optimizing for measured proxies rather than intended goals. This terminological diversity creates friction: comparing findings across studies requires mapping between frameworks, and comprehensive risk coverage requires consulting multiple taxonomies that use different organizing principles. This paper addresses this challenge by creating a comprehensive catalog of AI risks. We systematically analyzed every major AI risk framework published to date-74 frameworks containing 1,725 distinct risks-and organized them into a unified system. Our two classification systems reveal important patterns: contrary to common assumptions, human decisions cause nearly as many AI risks (38%) as the AI systems themselves (42%). The work provides practical tools for anyone working on AI safety, from developers conducting risk assessments to policymakers writing regulations to auditors evaluating AI systems. By establishing a common reference point, this repository creates the foundation for more coordinated and comprehensive approaches to managing AI's risks while realizing its benefits.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A consolidated database from 74 frameworks with dual taxonomies and a 38/42 human-vs-AI split, but the split rests on interpretive mappings whose consistency is not yet shown.

read the letter

The main deliverable is a single repository that pulls 74 AI risk frameworks and 1,725 risks into one place, then layers two new classification systems on top. The standout number is that human decisions account for 38% of the risks while the AI systems themselves account for 42%. That quantitative breakdown is new and directly challenges the usual emphasis on purely technical failure modes. The practical output is a shared reference that should cut down on the constant translation work between papers that use different terms for the same issue. Anyone doing risk assessments, writing policy, or auditing systems now has a common starting point instead of stitching together separate taxonomies. The scale of the review is the real contribution here; it is not just another framework but an attempt to organize everything that already exists. The soft spot is exactly the one the stress-test flags. The 38/42 split requires assigning every risk to one of two cause buckets, and the abstract gives no numbers on inter-annotator agreement, no decision rules for mixed cases, and no sensitivity checks. If even a modest fraction of the 1,725 items sit near the boundary, the percentages could shift. The methods section will need to show how frameworks were chosen and how duplicates were removed before the coverage claim can be taken at face value. This is the sort of reference paper that AI governance and safety groups will keep on their desks. It does not advance new theory or run new experiments, but the synthesis is useful enough that it deserves referee scrutiny so the classification rules and data can be examined before the repository gets widely adopted.

Referee Report

1 major / 1 minor

Summary. The paper conducts a systematic meta-review of 74 AI risk frameworks containing 1,725 distinct risks. It organizes these into a unified repository using two classification systems and reports that human decisions cause 38% of risks while AI systems cause 42%, providing a common reference point for risk assessment, regulation, and auditing.

Significance. If the mapping process is validated, the repository supplies a practical, shared taxonomy and database that could reduce terminological friction across studies and support coordinated AI risk management. The finding that human decisions contribute nearly as much as technical AI properties offers a concrete, falsifiable pattern that shifts focus toward governance and decision-making interventions.

major comments (1)

[Methods] Methods section (classification and coding procedure): the paper does not report inter-annotator agreement, explicit decision rules for hybrid or borderline risks, or sensitivity analyses on the human-vs-AI cause attribution. This directly undermines confidence in the 38%/42% split, as the headline quantitative claim rests on the reliability of assigning each of the 1,725 risks to one of the two cause categories.

minor comments (1)

[Abstract] Abstract and §2: the claim to have analyzed 'every major AI risk framework published to date' would be strengthened by an explicit search protocol, inclusion/exclusion criteria, and date cutoff.

Simulated Author's Rebuttal

1 responses · 1 unresolved

We thank the referee for this constructive comment on methodological transparency. We address the points below and will revise the manuscript to strengthen the presentation of our classification procedures.

read point-by-point responses

Referee: [Methods] Methods section (classification and coding procedure): the paper does not report inter-annotator agreement, explicit decision rules for hybrid or borderline risks, or sensitivity analyses on the human-vs-AI cause attribution. This directly undermines confidence in the 38%/42% split, as the headline quantitative claim rests on the reliability of assigning each of the 1,725 risks to one of the two cause categories.

Authors: We agree that additional detail on the classification process is warranted. In the revised version we will expand the Methods section to provide explicit decision rules for assigning risks to human-decision, AI-system, or hybrid categories, including concrete criteria and examples for handling borderline cases. We will also add a sensitivity analysis that varies the attribution thresholds and reports the resulting range for the 38%/42% split. The original coding was performed through iterative discussion and consensus among the core author team rather than by independent annotators; therefore we cannot report inter-annotator agreement statistics. We will state this limitation clearly and note its implications for the quantitative claims. revision: partial

standing simulated objections not resolved

Inter-annotator agreement metrics for the human-vs-AI cause attribution, as the coding was not performed by separate independent annotators

Circularity Check

0 steps flagged

No circularity: descriptive taxonomy from external sources

full rationale

The paper aggregates 1,725 risks from 74 independently published external frameworks and applies a new dual classification system to produce descriptive statistics such as the 38%/42% human-vs-AI split. No equations, fitted parameters, or self-citations are used to derive these figures; the percentages are direct outputs of the classification applied to sourced data. The process is an aggregation and reorganization exercise with no self-definitional loops or reductions to prior author work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

This is a meta-review and taxonomy paper whose central contribution is organizational. The primary assumption is the completeness and representativeness of the 74-framework sample; no free parameters or invented entities are introduced.

axioms (1)

domain assumption The 74 frameworks analyzed constitute every major AI risk framework published to date.
Stated directly in the abstract as the scope of the systematic analysis.

pith-pipeline@v0.9.0 · 5831 in / 1270 out tokens · 39206 ms · 2026-05-23T21:54:31.041680+00:00 · methodology

discussion (0)

Forward citations

Cited by 6 Pith papers

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

Culturally Aware GenAI Risks for Youth: Perspectives from Youth, Parents, and Teachers in a Non-Western Context
cs.HC 2026-04 unverdicted novelty 7.0

Mixed-methods research in Saudi Arabia reveals that GenAI use by youth creates culturally specific privacy and safety risks tied to family honor and shared accounts, requiring context-sensitive design.
To Build or Not to Build? Factors that Lead to Non-Development or Abandonment of AI Systems
cs.CY 2026-04 unverdicted novelty 6.0

A scoping review and empirical analysis produce a six-category taxonomy of factors driving AI non-development and abandonment, showing that practical issues like resource limits and organizational dynamics often outwe...
The 2025 AI Agent Index: Documenting Technical and Safety Features of Deployed Agentic AI Systems
cs.CY 2026-02 accept novelty 6.0

The 2025 AI Agent Index catalogs technical and safety details for 30 deployed AI agents and finds low developer transparency on safety, evaluations, and societal impacts.
A Closer Look at the Existing Risks of Generative AI: Mapping the Who, What, and How of Real-World Incidents
cs.CY 2025-05 unverdicted novelty 6.0

Analysis of 499 generative AI incidents shows use-related failures predominate and frequently harm non-users, producing a distinct risk profile from traditional AI.
What People See (and Miss) About Generative AI Risks: Perceptions of Failures, Risks, and Who Should Address Them
cs.HC 2026-04 unverdicted novelty 4.0

A validated survey instrument grounded in real GenAI incidents reveals public perceptions of failure modes, risks, and stakeholder responsibilities, showing potential for guiding AI literacy efforts.
Brainrot: Deskilling and Addiction are Overlooked AI Risks
cs.CY 2026-05 unverdicted novelty 3.0

AI safety literature overlooks cognitive deskilling and addiction risks from generative AI despite public concern about them.

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages · cited by 6 Pith papers · 1 internal anchor

[1]

AI Verify Foundation. (2023). Summary Report for Binary Classiﬁcation Model of Credit Risk . AI Verify Foundation. Allianz Global Corporate & Security. (2018). The rise of artiﬁcial intelligence: future outlooks and emerging risks . Allianz Global Corporate & Specialty SE . https://commercial.allianz.com/news-and-insights/reports/the-rise-of-artiﬁcial-int...

work page doi:10.1016/j.ress.2011.11.006 2023
[2]

other speciﬁed disorders due to addictive behaviors

https://doi.org/10.1186/s13643-024-02502-7 Bourget, D., & Chalmers, D. (n.d.). PHILOSOPHERS ON PHILOSOPHY: THE 2020 PHILPAPERS SURVEY . Brand, M., Rumpf, H.-J., Demetrovics, Z., MÜller, A., Stark, R., King, D. L., Goudriaan, A. E., Mann, K., Trotzke, P., Fineberg, N. A., Chamberlain, S. R., Kraus, S. W., Wegmann, E., Billieux, J., & Potenza, M. N. (2020)....

work page doi:10.1186/s13643-024-02502-7 2020
[3]

https://doi.org/10.1007/s10648-024-09862-5 Carroll, C., Booth, A., & Cooper, K. (2011). A worked example of “best ﬁt” framework synthesis: a systematic review of views concerning the taking of some potential chemopreventive agents. BMC Medical Research Methodology , 11 ,

work page doi:10.1007/s10648-024-09862-5 2011
[4]

https://doi.org/10.1186/1471-2288-11-29 Carroll, C., Booth, A., Leaviss, J., & Rick, J. (2013). “Best ﬁt” framework synthesis: reﬁning the method. BMC Medical Research Methodology , 13 ,

work page doi:10.1186/1471-2288-11-29 2013
[5]

A., Suárez-Betancur, N., Vega, L

https://doi.org/10.1186/1471-2288-13-37 Castaño-Pulgarín, S. A., Suárez-Betancur, N., Vega, L. M. T., & López, H. M. H. (2021). Internet, social media and online hate speech. Systematic review. Aggression and Violent Behavior , 58 , 101608. https://doi.org/10.1016/j.avb.2021.101608 Center for AI Safety. (2023, May 30). Statement on AI Risk . Center for AI...

work page doi:10.1186/1471-2288-13-37 2021
[6]

Journal of Clinical Epidemiology , 91 , 23–30

Introduction-the why, what, when, and how. Journal of Clinical Epidemiology , 91 , 23–30. https://doi.org/10.1016/j.jclinepi.2017.08.010 Epoch AI. (2024). Data on Notable AI Models [Dataset]. https://epochai.org/data/notable-ai-models European Commission. (COM/2021/206ﬁnal, 2021/0106(COD), 2021). Proposal For a Regulation of The European Parliament andof ...

work page doi:10.1016/j.jclinepi.2017.08.010 2017
[7]

https://doi.org/10.1186/s13643-023-02257-7 Ferrara, E. (2024). GenAI against humanity: nefarious applications of generative artiﬁcial intelligence and large language models. Journal of Computational Social Science , 7 (1), 549–569. https://doi.org/10.1007/s42001-024-00250-1 Francken, J. C., Beerendonk, L., Molenaar, D., Fahrenfort, J. J., Kiverstein, J. D...

work page doi:10.1186/s13643-023-02257-7 2024
[8]

F., … Vanschoren, J

https://doi.org/10.1186/s13643-019-1222-2 Ghosh, S., Frase, H., Williams, A., Luger, S., Röttger, P., Barez, F., McGregor, S., Fricklas, K., Kumar, M., Feuillade--Montixi, Q., Bollacker, K., Friedrich, F., Tsang, R., Vidgen, B., Parrish, A., Knotz, C., Presani, E., Bennion, J., Boston, M. F., … Vanschoren, J. (2025). AILUMINATE: Introducing v1.0 of the AI...

work page doi:10.1186/s13643-019-1222-2 2025
[9]

A., Ventura, C

https://doi.org/10.1186/2046-4053-1-10 Kilian, K. A., Ventura, C. J., & Bailey, M. M. (2023). Examining the differential risk from high-level artiﬁcial intelligence and the question of control. Futures , 151 (103182), 103182. https://doi.org/10.1016/j.futures.2023.103182 Kuhn, T. S. (1997). The structure of scientiﬁc revolutions (Vol. 962). University of ...

work page doi:10.1186/2046-4053-1-10 2046
[10]

Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment

86 https://doi.org/10.3390/su15053939 Liu, Y., Yao, Y., Ton, J.-F., Zhang, X., Guo, R., Cheng, H., Klochkov, Y., Tauﬁq, M. F., & Li, H. (2023). Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models’ Alignment. In arXiv [cs.AI] . arXiv. http://arxiv.org/abs/2308.05374 Marchal, N., Xu, R., Elasmar, R., Gabriel, I., Goldberg, B., & Is...

work page internal anchor Pith review Pith/arXiv arXiv doi:10.3390/su15053939 2023
[11]

C., Shoham, Y., Wald, R., & Clark, J

https://doi.org/10.12688/wellcomeopenres.19363.2 Maslej, N., Fattorini, L., Perrault, R., Parli, V., Reuel, A., Brynjolfsson, E., Etchemendy, J., Ligett, K., Lyons, T., Manyika, J., Niebles, J. C., Shoham, Y., Wald, R., & Clark, J. (2024). The AI Index 2024 Annual Report . AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, ...

work page doi:10.12688/wellcomeopenres.19363.2 2024
[12]

M., Silveira, F

https://doi.org/10.1186/s13012-015-0242-0 Paes, V. M., Silveira, F. F., & Akkari, A. C. S. (2023). Social impacts of artiﬁcial intelligence and mitigation recommendations: An exploratory study. In Proceedings of the 7th Brazilian Technology Symposium (BTSym’21) (pp. 521–528). Springer International Publishing. https://doi.org/10.1007/978-3-031-04435-9_54 ...

work page doi:10.1186/s13012-015-0242-0 2023
[13]

https://doi.org/10.3390/app12084054 Schnitzer, R., Hapfelmeier, A., Gaube, S., & Zillner, S. (2023). AI Hazard Management: A framework for the systematic management of root causes for AI risks. In arXiv [cs.LG] . arXiv. http://arxiv.org/abs/2310.16727 Sharma, S. (2024). Beneﬁts or concerns of AI: A multistakeholder responsibility. Futures , 157 , 103328. ...

work page doi:10.3390/app12084054 2023
[14]

C., Langlois, E

90 https://doi.org/10.1186/1471-2288-8-45 Tricco, A. C., Langlois, E. V., & Straus, S. E. (2017). Rapid reviews to strengthen health policy and systems: a practical guide . World Health Organization. https://ahpsr.who.int/publications/i/item/2017-08-10-rapid-reviews-to-strengthen-health-polic y-and-systems-a-practical-guide UK Department for Science, Inno...

work page doi:10.1186/1471-2288-8-45 2017

[1] [1]

AI Verify Foundation. (2023). Summary Report for Binary Classiﬁcation Model of Credit Risk . AI Verify Foundation. Allianz Global Corporate & Security. (2018). The rise of artiﬁcial intelligence: future outlooks and emerging risks . Allianz Global Corporate & Specialty SE . https://commercial.allianz.com/news-and-insights/reports/the-rise-of-artiﬁcial-int...

work page doi:10.1016/j.ress.2011.11.006 2023

[2] [2]

other speciﬁed disorders due to addictive behaviors

https://doi.org/10.1186/s13643-024-02502-7 Bourget, D., & Chalmers, D. (n.d.). PHILOSOPHERS ON PHILOSOPHY: THE 2020 PHILPAPERS SURVEY . Brand, M., Rumpf, H.-J., Demetrovics, Z., MÜller, A., Stark, R., King, D. L., Goudriaan, A. E., Mann, K., Trotzke, P., Fineberg, N. A., Chamberlain, S. R., Kraus, S. W., Wegmann, E., Billieux, J., & Potenza, M. N. (2020)....

work page doi:10.1186/s13643-024-02502-7 2020

[3] [3]

https://doi.org/10.1007/s10648-024-09862-5 Carroll, C., Booth, A., & Cooper, K. (2011). A worked example of “best ﬁt” framework synthesis: a systematic review of views concerning the taking of some potential chemopreventive agents. BMC Medical Research Methodology , 11 ,

work page doi:10.1007/s10648-024-09862-5 2011

[4] [4]

https://doi.org/10.1186/1471-2288-11-29 Carroll, C., Booth, A., Leaviss, J., & Rick, J. (2013). “Best ﬁt” framework synthesis: reﬁning the method. BMC Medical Research Methodology , 13 ,

work page doi:10.1186/1471-2288-11-29 2013

[5] [5]

A., Suárez-Betancur, N., Vega, L

https://doi.org/10.1186/1471-2288-13-37 Castaño-Pulgarín, S. A., Suárez-Betancur, N., Vega, L. M. T., & López, H. M. H. (2021). Internet, social media and online hate speech. Systematic review. Aggression and Violent Behavior , 58 , 101608. https://doi.org/10.1016/j.avb.2021.101608 Center for AI Safety. (2023, May 30). Statement on AI Risk . Center for AI...

work page doi:10.1186/1471-2288-13-37 2021

[6] [6]

Journal of Clinical Epidemiology , 91 , 23–30

Introduction-the why, what, when, and how. Journal of Clinical Epidemiology , 91 , 23–30. https://doi.org/10.1016/j.jclinepi.2017.08.010 Epoch AI. (2024). Data on Notable AI Models [Dataset]. https://epochai.org/data/notable-ai-models European Commission. (COM/2021/206ﬁnal, 2021/0106(COD), 2021). Proposal For a Regulation of The European Parliament andof ...

work page doi:10.1016/j.jclinepi.2017.08.010 2017

[7] [7]

https://doi.org/10.1186/s13643-023-02257-7 Ferrara, E. (2024). GenAI against humanity: nefarious applications of generative artiﬁcial intelligence and large language models. Journal of Computational Social Science , 7 (1), 549–569. https://doi.org/10.1007/s42001-024-00250-1 Francken, J. C., Beerendonk, L., Molenaar, D., Fahrenfort, J. J., Kiverstein, J. D...

work page doi:10.1186/s13643-023-02257-7 2024

[8] [8]

F., … Vanschoren, J

https://doi.org/10.1186/s13643-019-1222-2 Ghosh, S., Frase, H., Williams, A., Luger, S., Röttger, P., Barez, F., McGregor, S., Fricklas, K., Kumar, M., Feuillade--Montixi, Q., Bollacker, K., Friedrich, F., Tsang, R., Vidgen, B., Parrish, A., Knotz, C., Presani, E., Bennion, J., Boston, M. F., … Vanschoren, J. (2025). AILUMINATE: Introducing v1.0 of the AI...

work page doi:10.1186/s13643-019-1222-2 2025

[9] [9]

A., Ventura, C

https://doi.org/10.1186/2046-4053-1-10 Kilian, K. A., Ventura, C. J., & Bailey, M. M. (2023). Examining the differential risk from high-level artiﬁcial intelligence and the question of control. Futures , 151 (103182), 103182. https://doi.org/10.1016/j.futures.2023.103182 Kuhn, T. S. (1997). The structure of scientiﬁc revolutions (Vol. 962). University of ...

work page doi:10.1186/2046-4053-1-10 2046

[10] [10]

Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment

86 https://doi.org/10.3390/su15053939 Liu, Y., Yao, Y., Ton, J.-F., Zhang, X., Guo, R., Cheng, H., Klochkov, Y., Tauﬁq, M. F., & Li, H. (2023). Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models’ Alignment. In arXiv [cs.AI] . arXiv. http://arxiv.org/abs/2308.05374 Marchal, N., Xu, R., Elasmar, R., Gabriel, I., Goldberg, B., & Is...

work page internal anchor Pith review Pith/arXiv arXiv doi:10.3390/su15053939 2023

[11] [11]

C., Shoham, Y., Wald, R., & Clark, J

https://doi.org/10.12688/wellcomeopenres.19363.2 Maslej, N., Fattorini, L., Perrault, R., Parli, V., Reuel, A., Brynjolfsson, E., Etchemendy, J., Ligett, K., Lyons, T., Manyika, J., Niebles, J. C., Shoham, Y., Wald, R., & Clark, J. (2024). The AI Index 2024 Annual Report . AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, ...

work page doi:10.12688/wellcomeopenres.19363.2 2024

[12] [12]

M., Silveira, F

https://doi.org/10.1186/s13012-015-0242-0 Paes, V. M., Silveira, F. F., & Akkari, A. C. S. (2023). Social impacts of artiﬁcial intelligence and mitigation recommendations: An exploratory study. In Proceedings of the 7th Brazilian Technology Symposium (BTSym’21) (pp. 521–528). Springer International Publishing. https://doi.org/10.1007/978-3-031-04435-9_54 ...

work page doi:10.1186/s13012-015-0242-0 2023

[13] [13]

https://doi.org/10.3390/app12084054 Schnitzer, R., Hapfelmeier, A., Gaube, S., & Zillner, S. (2023). AI Hazard Management: A framework for the systematic management of root causes for AI risks. In arXiv [cs.LG] . arXiv. http://arxiv.org/abs/2310.16727 Sharma, S. (2024). Beneﬁts or concerns of AI: A multistakeholder responsibility. Futures , 157 , 103328. ...

work page doi:10.3390/app12084054 2023

[14] [14]

C., Langlois, E

90 https://doi.org/10.1186/1471-2288-8-45 Tricco, A. C., Langlois, E. V., & Straus, S. E. (2017). Rapid reviews to strengthen health policy and systems: a practical guide . World Health Organization. https://ahpsr.who.int/publications/i/item/2017-08-10-rapid-reviews-to-strengthen-health-polic y-and-systems-a-practical-guide UK Department for Science, Inno...

work page doi:10.1186/1471-2288-8-45 2017