Pith. sign in

REVIEW 4 major objections 7 minor 89 references

Recourse, Repair, Reparation, & Prevention: A Stakeholder Analysis of AI Supply Chains

T0 review · 4 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that every response to harm in an AI supply chain takes one of four forms—recourse, repair, reparation, or prevention—and that achieving any of them is gated by consensus among the necessary actors and by whether the…

desk verdict A useful conceptual framework for AI supply chain redress, but the two-factor claim needs to be reframed as a heuristic before the paper can carry the weight it asks for. read the letter →

arxiv 2507.02648 v1 pith:OZ4SKAQS submitted 2025-07-03 cs.CY

Aspen K. Hopkins , Isabella Struckman , Kevin Klyman , Susan S. Silbey This is my paper
classification cs.CY
keywords AIsupplychainsstakeholderanalysisredresstypologyrecourserepairreparationpreventionmarketstructure
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

AI systems are usually assembled from layered networks of outsourced models, data, and infrastructure—what the paper calls AI supply chains. The paper tries to establish that when such a chain produces harm, only four kinds of response exist: recourse (stopping the harm), repair (correcting it), reparation (compensating for it), and prevention (mitigating it in advance). It further claims that whether any of these is actually achieved is determined by two conditions: consensus, meaning the necessary actors agree or can be compelled to accept the remedy, and achievability, meaning the remedy is physically, technically, or legally possible. The authors then apply this typology to a healthcare supply chain—an EHR platform, a voice-to-text model, a sepsis model, and hospitals—under vertical integration, horizontal integration, and free-market competition, arguing that market structure shapes who has power to demand or block redress. A sympathetic reader would care because the typology gives regulators and designers a common language for predicting when harms will go unaddressed.

What carries the argument

The load-bearing object is a two-part test for whether redress exists: consensus, defined as the necessary actors agreeing or being effectively compelled to accept a proposed remedy, and achievability, defined as the remedy being physically, technically, or legally possible given the parties' and systems' limitations. Around this test, the paper builds a four-way typology of redress—recourse, repair, reparation, prevention—and applies it through three stylized market structures (vertical integration, horizontal integration, and free market) that determine the bargaining power and informational position of each stakeholder. The test does the explanatory work: it converts the question 'who is liable?' into the question 'who can say yes or no, and can the fix actually be done?'

What would settle it

One concrete test: in any real AI supply chain harm, record whether the responsible parties agreed to a remedy and whether the remedy was physically, technically, and legally possible, then observe whether redress occurred. A case where both conditions held and yet no redress happened—for instance, hospitals and model providers all agree to revert an update and the revert is technically simple, but patients still receive no reparation because they lack legal standing—would show the two conditions are not sufficient.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the sprawling, loosely coupled networks that produce AI systems concentrate power in ways that determine which remedies are available after a harm, regardless of who is morally or legally at fault. The paper identifies six stakeholder roles—infrastructure, data, model, intermediary, user-facing, and users/consumers—and catalogues mechanisms of harm that include familiar AI risks such as false content and biased decisions plus supply-chain-specific ones such as diffused responsibility, reduced optionality, and homogenization. It argues that redress succeeds only when there is consensus (agreement or effective compulsion) and achievability (physical, technical, or legal possibility). Across the three stylized markets, the paper predicts that vertical integration lets the integrated provider set repair priorities and blocks outside recourse, horizontal integration can create a monitoring intermediary that enables prevention, and free markets make recourse and reparation easier but repairs harder without transparency.

Load-bearing premise

The model rests on the premise that whether any redress is achieved is fully explained by consensus and achievability, with all other factors—legal standing, information asymmetry, reputation, emotion—operating only through those two gates; if power can travel through channels outside those two, the case-study predictions can fail.

Editorial extensions

If this is right

  • In vertically integrated AI supply chains, redress is dominated by the integrated provider: reparations are hard to extract, and outside stakeholders like hospitals can rarely opt out of a single component.
  • In horizontally integrated chains, an independent intermediary such as an EHR vendor can observe both upstream model changes and downstream effects, making repair and prevention more tractable than in a monolith.
  • In free markets, affected parties can stop harm by switching providers and can often obtain refunds or discounts, but repairs may fail if no one can see across the chain to find the faulty component.
  • The two-gate test implies that regulation works by changing consensus: a legal mandate can compel acceptance of a remedy that voluntary negotiation would never produce.
  • The typology provides a shared vocabulary for tracing why harms persist even when every actor agrees a fix is technically possible.

Reading between the lines

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

  • If consensus can be manufactured by compulsion, then the paper's two-condition account points to liability rules and regulatory orders as the main levers for redress; the paper itself does not specify which institutions are best positioned to compel consensus.
  • The two conditions may not be independent: achievability often depends on information held by the counterparty, so disputes about traceability can disguise themselves as technical impossibility.
  • A natural test is to code documented incidents (such as the CrowdStrike outage that the paper names) against the four redress forms and see whether observed outcomes track consensus and achievability rather than legal standing, information asymmetry, or reputational pressure.
  • The paper's tables imply that prevention is the most fragile form of redress because it requires buy-in from parties who have not yet been harmed; the paper does not state this as an explicit hypothesis.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The paper develops a stakeholder analysis of AI supply chains (AISCs), identifying six stakeholder roles and nine mechanisms of harm, then introduces a typology of four forms of redress—recourse, repair, reparation, and prevention. The central analytic claim is that whether redress is achieved is determined by two factors: consensus and achievability. This framework is applied to a hypothetical healthcare AISC in three stylized market structures—vertical integration, horizontal integration, and a free market—yielding comparative predictions about which redress routes are available to each stakeholder. The paper is explicitly conceptual: it states in its limitations that it does not employ formal empirical methods and that the typology is based on conceptual synthesis from existing scholarship.

Significance. If the two-factor model could be operationalized, the paper would offer a useful organizing device for thinking about redress in AI supply chains. The stakeholder mapping and the redress typology bring together literatures from stakeholder theory, supply chain management, law, and ML fairness in a way that is genuinely cross-disciplinary. The paper is transparent about its limitations, which is a strength, and the healthcare case offers a concrete, comparable setting for reasoning about market structure. However, the central claim—that consensus and achievability determine redress—is asserted rather than derived or tested, and the predictions in Tables 2 and 3 depend on auxiliary assumptions that are not among the stated model parameters. The contribution is best understood as a conceptual framework in need of empirical grounding and sharper formalization rather than as an established predictive model.

major comments (4)
  1. [Section 5] The claim that "[w]hether a redress is achieved in an AI supply chain is determined by two considerations" is a strong sufficiency assertion, but the manuscript does not derive it from prior theory and does not specify how consensus and achievability are measured, operationalized, or weighted. As written, any observed absence of redress can be post hoc attributed to a lack of consensus or achievability, making the model difficult to falsify. I recommend either specifying independent criteria for assessing consensus and achievability before observing an outcome, or reframing the claim as a heuristic rather than a deterministic statement.
  2. [Section 7, Tables 2-3] The comparative predictions in Tables 2 and 3 (e.g., "Reparation: Achievable Yes, through litigation. Liability is provable" under horizontal integration; "Repair: Maybe" in the free market) rely on auxiliary assumptions about traceability, contract terms, legal standing, information asymmetry, and switching costs that are not part of the stated two-factor model. The narratives in Section 7 mention these factors, but they are not encoded in the consensus/achievability determinants, leaving the reader unable to distinguish which predictions follow from the model and which are imported from the scenario description. The paper should either make these auxiliary assumptions explicit and variable or restrict the model's claims accordingly.
  3. [Table 3] The Reparation column in Table 3 marks reparation as "Irrelevant" for the fiscal-loss harm incurred by (C) across all three market structures, but the case-study narrative in Section 7 states that (C) "may look for reparations given income or reputational losses." This is a direct inconsistency between the table and the text. The table should be revised to align with the scenario, or the text should explain why reparation is excluded.
  4. [Section 8 (Limitations)] The paper concedes that stakeholder analysis "often under-theorizes topics like resistance, marginality, or contestation" and that it "assum[es] consensus is ideal." This concession directly challenges the sufficiency of the consensus/achievability model, because channels such as protest, non-participation, or contestation can shape redress outcomes even when consensus and achievability are present in the authors' sense. The claim that the case studies mitigate this issue is not supported: the healthcare scenarios are hypothetical and do not engage these channels. The model's scope should be narrowed, or this limitation should be integrated into the Central framing of Section 5.
minor comments (7)
  1. [Author affiliations] "Massachussetts" is misspelled on the author affiliation lines.
  2. [Section 1.1] "neccessary" should be "necessary" in the sentence beginning "common actions neccessary in addressing."
  3. [Section 4.2] "downstram" should be "downstream" in "struggle to recognize biasing effects across the AISC."
  4. [Section 4.9] "propegating" should be "propagating" in "further concentrates design choices, propegating this similarity."
  5. [Section 7] "achievabile" should be "achievable" and "shapping" should be "shaping" in the sentence "market dynamics shapping the AI supply chain."
  6. [Section 7.1] The sentence "interoperability between components does not non-existent" appears to be missing words and should be revised to something like "interoperability between components is not non-existent."
  7. [Section 7.3] "best practies" should be "best practices" in the final paragraph of the subsection.

Circularity Check

0 steps flagged · score 1.0 of 10

Typology is definitional but not derived from fitted inputs or the authors' own prior results; no significant circularity.

full rationale

The paper's central contribution is a conceptual typology of redress (recourse, repair, reparation, prevention) and a two-factor heuristic (consensus, achievability). The typology is explicitly presented as a 'logical synthesis produced through conceptual comparison and alignment' (Section 5), not as a result derived from fitted data, equations, or the authors' prior theorems. The case-study tables (Tables 2-3) apply this typology qualitatively to a stylized healthcare AISC; their 'Achievable: Yes/Maybe' entries are illustrative assertions rather than numerically forced predictions. The authors' self-citations ([6] dataset, [38] AI supply chains framework) support background stakeholder mapping and the definition of AISC, but they do not determine the redress typology or the consensus/achievability condition. The limitations section candidly states that 'the stakeholder roles and redress typology are based on conceptual synthesis from existing published scholarship rather than through stakeholder accounts or case-based evidence,' which confirms the claims are asserted rather than circularly derived. A possible concern is that the two-factor model is broad enough that outcomes could be rationalized ex post, and the tables also invoke power and urgency beyond the two stated factors; however, that is a falsifiability or operationalization issue, not a circular reduction of the paper's predictions to its inputs. No equation or definition makes a 'prediction' identical to a fitted parameter or to a same-author uniqueness theorem, so the derivation chain is not circular.

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

The paper introduces no new entities, particles, or formal constructs. It relies on existing concepts from stakeholder theory, organizational theory, and the authors' prior work on AI supply chains. The main assumptions are about the nature of redress and the classification of market structures.

assumptions (4)
  • domain assumption Redress is governed by consensus and achievability, and these two factors are sufficient to explain whether a response occurs.
    Stated in Section 5 as the determinant of redress in AI supply chains. Both terms are defined conceptually, but the claim that they are the sole or primary determinants is asserted, not derived.
  • domain assumption The four forms of redress (recourse, repair, reparation, prevention) are exhaustive and mutually distinct.
    The typology is introduced in the abstract and Section 5 as a complete categorization, but the paper does not argue for exhaustiveness or demonstrate that no other form of redress exists.
  • domain assumption Market structures can be meaningfully classified as vertical integration, horizontal integration, or free market, and these categories are stable enough to predict redress outcomes.
    The model in Section 6 and the case study in Section 7 rely on these stylized categories, but real AI supply chains often mix structures and shift over time, which the paper acknowledges only in passing.
  • domain assumption AI supply chains are the result of human agency and can be re-designed to account for harms.
    This is the motivating premise in the Introduction ('AI supply chains are invented and implemented by humans with agency'), which supports the paper's normative goal but is not empirically demonstrated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Recourse, Repair, Reparation, & Prevention: A Stakeholder Analysis of AI Supply Chains." pith.science (2026). https://pith.science/paper/OZ4SKAQS

@misc{pith2026250702648,
  author       = {Pith},
  title        = {Pith review of: Recourse, Repair, Reparation, & Prevention: A Stakeholder Analysis of AI Supply Chains},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OZ4SKAQS}},
  note         = {Machine review of arXiv:2507.02648}
}
abstract

The AI industry is exploding in popularity, with increasing attention to potential harms and unwanted consequences. In the current digital ecosystem, AI deployments are often the product of AI supply chains (AISC): networks of outsourced models, data, and tooling through which multiple entities contribute to AI development and distribution. AI supply chains lack the modularity, redundancies, or conventional supply chain practices that enable identification, isolation, and easy correction of failures, exacerbating the already difficult processes of responding to ML-generated harms. As the stakeholders participating in and impacted by AISCs have scaled and diversified, so too have the risks they face. In this stakeholder analysis of AI supply chains, we consider who participates in AISCs, what harms they face, where sources of harm lie, and how market dynamics and power differentials inform the type and probability of remedies. Because AI supply chains are purposely invented and implemented, they may be designed to account for, rather than ignore, the complexities, consequences, and risks of deploying AI systems. To enable responsible design and management of AISCs, we offer a typology of responses to AISC-induced harms: recourse, repair, reparation or prevention. We apply this typology to stakeholders participating in a health-care AISC across three stylized markets $\unicode{x2013}$ vertical integration, horizontal integration, free market $\unicode{x2013}$ to illustrate how stakeholder positioning and power within an AISC may shape responses to an experienced harm.

Figures

Figures reproduced from arXiv: 2507.02648 by the authors.

Figure 1
Figure 1. Stakeholders & redress typology in AISCs. (A) An overview of stakeholder roles in an AI supply chain, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. General AISC and sepsis model AISC are shown under vertical integration, horizontal integration, and the [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

89 extracted references · 69 canonical work pages

  1. [1]

    Andrew Abbott. 1988. The System of Professions: An Essay on the Division of Expert Labor. University of Chicago Press, Chicago

    1988

  2. [2]

    NIST AI. 2023. Artificial Intelligence Risk Management Framework (AI RMF 1.0)

    2023

  3. [3]

    George A. Akerlof. 1970. The Market for "Lemons": Quality Uncertainty and the Market Mechanism. The Quarterly Journal of Economics 84, 3 (1970), 488–500

    1970

  4. [4]

    Athanasios Andreou, Oana Goga, and Patrick Loiseau. 2017. Identity vs. attribute disclosure risks for users with multiple social profiles. In Proceedings of the 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2017 . IEEE/ACM

    2017

  5. [5]

    Maria Cristina Arcuri, Gino Gandolfi, and Ivan Russo. 2023. Does fake news impact stock returns? Evidence from US and EU stock markets. Journal of Economics and Business 125 (2023), 106130

    2023

  6. [6]

    Aspen Hopkins and Isabella Struckman, Aleksander Madry, and Luis Videgaray. 2024. AIaaS Supply Chains Dataset Repository. https://github. com/MadryLab/AIaaS_Supply_Chains GitHub repository

    2024

  7. [7]

    Rishi Bommasani, Kathleen A Creel, Ananya Kumar, Dan Jurafsky, and Percy S Liang. 2022. Picking on the same person: Does algorithmic mono- culture lead to outcome homogenization? Advances in Neural Information Processing Systems 35 (2022), 3663–3678

    2022

  8. [8]

    Liao, Kathleen A

    Rishi Bommasani, Dilara Soylu, Thomas I. Liao, Kathleen A. Creel, and Percy Liang. 2023. Ecosystem Graphs: The Social Footprint of Foundation Models. (March 2023). https://arxiv.org/abs/2303.15772

    arXiv 2023

Show all 89 references
  1. [9]

    Rishi Bommasani, Dilara Soylu, Thomas I Liao, Kathleen A Creel, and Percy Liang. 2024. Ecosystem Graphs: Documenting the Foundation Model Supply Chain. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, Vol. 7. 196–209

    2024

  2. [10]

    Shamik Chaudhuri, Kingshuk Dasgupta, Isaac Hepworth, Michael Le, Mark Lodato, Mihai Maruseac, Sarah Meiklejohn, Tehila Minkus, and Kara Olive. 2024. Securing the AI Software Supply Chain . Technical Report. Google. https://storage.googleapis.com/gweb-research2023- media/pubtoo...

    2024

  3. [11]

    Max Clarkson. 1994. A risk based model of stakeholder theory. In Proceed- ings of the second Toronto conference on stakeholder theory . 18–19

    1994

  4. [12]

    Jennifer Cobbe, Chris Norval, and Jatinder Singh. 2020. What Lies Beneath: Transparency in Online Service Supply Chains. Journal of Cyber Policy 5, 1 (2020), 65–93. doi:10.1080/23738871.2020.1719036

    2020

  5. [13]

    Jennifer Cobbe, Michael Veale, and Jatinder Singh. 2023. Understanding Accountability in Algorithmic Supply Chains. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency. ACM, 1186–

    2023

  6. [14]

    Feder Cooper, Shagun Jhaver, Eric Gilbert, et al

    A. Feder Cooper, Shagun Jhaver, Eric Gilbert, et al. 2022. Accountability in an algorithmic society: relationality, responsibility, and robustness in machine learning. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency. ACM, 214–227. doi:10...

    2022 doi

  7. [15]

    Yunkai Dang, Kaichen Huang, Jiahao Huo, Yibo Yan, Sirui Huang, Dongrui Liu, Mengxi Gao, Jie Zhang, Chen Qian, Kun Wang, et al. 2024. Explainable and interpretable multimodal large language models: A comprehensive survey. arXiv preprint arXiv:2412.02104 (2024)

    2024 arXiv

  8. [16]

    Pablo De Greiff. 2008. The handbook of reparations . Oxford University Press

    2008

  9. [17]

    Jerry Dischler. 2025. Empowering Businesses with AI in Google Workspace. https://workspace.google.com/blog/product-announcements/ empowering-businesses-with-ai?utm_campaign=article_email&utm_ content=article-14309&utm_medium=email&utm_source=sg Accessed: 2025-01-20

    2025

  10. [18]

    Thomas Donaldson and Lee E Preston. 1995. The stakeholder theory of the corporation: Concepts, evidence, and implications. Academy of management Review 20, 1 (1995), 65–91

    1995

  11. [19]

    Faizan Farooque. 2024. Microsoft Signs Deal with HarperCollins for AI Model Training. (November 20 2024). https://finance.yahoo.com/news/ microsoft-signs-deal-harpercollins-ai-161159188.html

    2024

  12. [20]

    Federal Communications Commission. 2021. FCC Fines Telemarketer $225 Million for Spoofed Robocalls. https://www.fcc.gov/document/fcc-fines- telemarketer-225-million-spoofed-robocalls Accessed: 2025-01-20

    2021

  13. [21]

    Federal Communications Commission. 2024. FCC SETTLES CASE AGAINST PROVIDER THAT TRANSMITTED SPOOFED AI-GENERATED ROBOCALLS FOR ELECTION INTERFERENCE IN NEW HAMPSHIRE. https://docs.fcc.gov/public/attachments/DOC-404951A1.pdf. Accessed: 2025-01-22

    2024

  14. [22]

    Carmen Fine. 2023. 10.5 Trillion Reasons Why We Need A United Response To Cyber Risk. https://www.forbes.com/sites/forbestechcouncil/ 2023/02/22/105-trillion-reasons-why-we-need-a-united-response-to- cyber-risk/?sh=3b3c46e83b0c. Accessed: 2025-01-22

    2023

  15. [23]

    Sara Fischer. 2025. OpenAI to fund four new Axios Local newsrooms. https://blog.google/products/news/associated-press-gemini-app/

    2025

  16. [24]

    Robert Edward Freeman. 1984. Strategic management: a stakeholder approach, Pitman. Boston, MA 521151740 (1984)

    1984

  17. [25]

    R Edward Freeman. 2010. Strategic management: A stakeholder approach . Cambridge university press

    2010

  18. [26]

    R Edward Freeman and William M Evan. 1990. Corporate governance: A stakeholder interpretation. Journal of behavioral economics 19, 4 (1990), 337–359

    1990

  19. [27]

    Jeff Frooman. 1999. Stakeholder influence strategies. Academy of manage- ment review 24, 2 (1999), 191–205

    1999

  20. [28]

    Andra C Ghent and Marianna Kudlyak. 2011. Recourse and residential mortgage default: evidence from US states. The Review of Financial Studies 24, 9 (2011), 3139–3186

    2011

  21. [29]

    Mark Gibney and Erik Roxstrom. 2001. The status of state apologies. Human Rights Quarterly 23, 4 (2001), 911–939

    2001

  22. [30]

    Andrew S Gold. 2014. A theory of redressive justice. University of Toronto Law Journal 64, 2 (2014), 159–205

    2014

  23. [31]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde- Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2020. Generative adversarial networks. Commun. ACM 63, 11 (2020), 139–144

    2020

  24. [32]

    Trudy Govier and Wilhelm Verwoerd. 2002. The promise and pitfalls of apology. Journal of social philosophy 33, 1 (2002)

    2002

  25. [33]

    Leon Griesch, Jack Rittelmeyer, and Kurt Sandkuhl. 2023. Towards AI as a Service for Small and Medium-Sized Enterprises (SME). In IFIP Working Conference on The Practice of Enterprise Modeling . Springer, 37–53

    2023

  26. [34]

    David Hamilton. 2024. Reddit strikes $60 M deal allowing Google to train AI models on its posts, unveils IPO plans. The Associated Press. Retrieved from https://apnews.com/article/google-reddit-ai-partnership- a7f131c7cb4225307134ef21d3c6a708. Accessed: 2025-01-22

    2024

  27. [35]

    P. J. Henning. 2014. Prosecutors often need secrecy to track down those behind cyberattacks. https://dealbook.nytimes.com/2014/01/28/adding- up-the-costs-of-data-breaches/. Accessed: 2025-01-22

    2014

  28. [36]

    Erik Hollnagel, David D Woods, and Nancy Leveson. 2006. Resilience engineering: Concepts and precepts . Ashgate Publishing, Ltd

    2006

  29. [37]

    Sungsoo Ray Hong, Jessica Hullman, and Enrico Bertini. 2020. Human factors in model interpretability: Industry practices, challenges, and needs. Proceedings of the ACM on Human-Computer Interaction 4, CSCW1 (2020), 1–26

    2020

  30. [38]

    Aspen Hopkins, Sarah H Cen, Andrew Ilyas, Isabella Struckman, Luis Videgaray, and Aleksander Mądry. 2025. AI Supply Chains: An Emerg- ing Ecosystem of AI Actors, Products, and Services. arXiv preprint arXiv:2504.20185 (2025)

    2025

  31. [39]

    Ruthanne Huising and Susan S Silbey. 2018. From nudge to culture and back again: Coalface governance in the regulated organization. Annual Review of Law and Social Science 14, 1 (2018), 91–114

    2018

  32. [40]

    Wiebke Hutiri, Orestis Papakyriakopoulos, and Alice Xiang. 2024. Not My Voice! A Taxonomy of Ethical and Safety Harms of Speech Generators. In The 2024 ACM Conference on Fairness, Accountability, and Transparency . Recourse, Repair, Reparation, & Prevention: A Stakeholder Anal...

    2024

  33. [41]

    Internet Crime Complaint Center (IC3). 2023. 2023 Elder Fraud Report. https://www.ic3.gov/AnnualReport/Reports/2023_IC3ElderFraudReport. pdf Accessed: 2025-01-20

    2023

  34. [42]

    Nathanael Jo, Kathleen Creel, Ashia Wilson, and Manish Raghavan. 2025. Homogeneous Algorithms Can Reduce Competition in Personalized Pric- ing. arXiv:2503.15634 [cs.GT] https://arxiv.org/abs/2503.15634

    2025 arXiv

  35. [43]

    TM Jones. 1999. Convergent Stakeholder Theory.Academy of Management Review 24, 2 (1999), 206–221

    1999

  36. [44]

    Tero Karppi and Kate Crawford. 2016. Social media, financial algorithms and the hack crash. Theory, culture & society 33, 1 (2016), 73–92

    2016

  37. [45]

    John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. 2024. A Watermark for Large Language Models. arXiv:2301.10226 [cs.LG] https://arxiv.org/abs/2301.10226

    2024 arXiv

  38. [46]

    Jon Kleinberg and Manish Raghavan. 2021. Algorithmic monoculture and social welfare. Proceedings of the National Academy of Sciences 118, 22 (2021), e2018340118

    2021

  39. [47]

    Sarah Kreps, R Miles McCain, and Miles Brundage. 2022. All the news that’s fit to fabricate: AI-generated text as a tool of media misinformation. Journal of experimental political science 9, 1 (2022), 104–117

    2022

  40. [48]

    Katherine Lee, A Feder Cooper, and James Grimmelmann. 2023. Talkin”Bout AI Generation: Copyright and the Generative-AI Supply Chain. arXiv preprint arXiv:2309.08133 (2023)

    2023 arXiv

  41. [49]

    Zachary C. Lipton. 2016. The mythos of model interpretability. arXiv preprint arXiv:1606.03490 (2016)

    2016 arXiv

  42. [50]

    Eric Justin Liu, Wonyoung So, Peko Hosoi, and Catherine D’Ignazio. 2024. Racial Steering by Large Language Models: A Prospective Audit of GPT-4 on Housing Recommendations. In Proceedings of the 4th ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization . 1–13

    2024

  43. [51]

    Alasdair MacIntyre. 1984. After Virtue: A Study in Moral Theory (2nd ed.). University of Notre Dame Press, Notre Dame, IN

    1984

  44. [52]

    MacKinnon

    Catharine A. MacKinnon. 1989. Toward a Feminist Theory of the State . Harvard University Press, Cambridge, MA

    1989

  45. [53]

    Martin M. Merener. 2012. Theoretical results on de-anonymization via linkage attacks. Trans. Data Priv. 5, 2 (2012), 377–402

    2012

  46. [54]

    Jacob Metcalf, Ranjit Singh, Emanuel Moss, Emnet Tafesse, and Eliza- beth Anne Watkins. 2023. Taking algorithms to courts: A relational approach to algorithmic accountability. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency . 1450–1462

    2023

  47. [55]

    Ronald K Mitchell, Bradley R Agle, and Donna J Wood. 1997. Toward a theory of stakeholder identification and salience: Defining the principle of who and what really counts. Academy of management review 22, 4 (1997), 853–886

    1997

  48. [56]

    Elting E Morison. 2016. Men, machines, and modern times . MIT press

    2016

  49. [57]

    Joseph Morrison. 2023. How I Broke into a Bank Account with an AI- Generated Voice. Vice (March 2023). https://www.vice.com/en/article/ how-i-broke-into-a-bank-account-with-an-ai-generated-voice

    2023

  50. [58]

    Mozilla Foundation. 2025. Public AI: Building Artificial Intelligence for the Public Good. https://assets.mofoprod.net/network/documents/Public_ AI_Mozilla.pdf. Accessed: 2025-01-22

    2025

  51. [59]

    Helen Nissenbaum. 1996. Accountability in a computerized society.Science and Engineering Ethics 2, 1 (1996), 25–42. doi:10.1007/BF02639315

    1996 doi

  52. [60]

    Douglass C. North. 1990. Institutions, Institutional Change, and Economic Performance. Cambridge University Press, Cambridge

    1990

  53. [61]

    Christine Oliver. 1991. Strategic responses to institutional processes. Academy of management review 16, 1 (1991), 145–179

    1991

  54. [62]

    1984.Normal Accidents: Living with High-Risk Technologies

    Charles Perrow. 1984.Normal Accidents: Living with High-Risk Technologies. Basic Books, New York

    1984

  55. [63]

    Salancik

    Jeffrey Pfeffer and Gerald R. Salancik. 1978. The External Control of Or- ganizations: A Resource Dependence Perspective . Harper and Row, New York

    1978

  56. [64]

    Robert Phillips. 1999. On stakeholder delimitation. Business and Society 38, 1 (1999), 32

    1999

  57. [65]

    John Rawls. 2020. A theory of justice: Revised edition . Harvard university press

    2020

  58. [66]

    Reuters. 2024. Meta Platforms to use Reuters news content in AI chatbot. (October 25 2024). https://www.reuters.com/technology/artificial- intelligence/meta-platforms-use-reuters-news-content-ai-chatbot- 2024-10-25/

    2024

  59. [67]

    Richard Rothstein. 2017. The color of law: A forgotten history of how our government segregated America. Liveright Publishing

    2017

  60. [68]

    Timothy J Rowley. 1997. Moving beyond dyadic ties: A network theory of stakeholder influences. Academy of management Review 22, 4 (1997), 887–910

    1997

  61. [69]

    Niki Saenz. 2024. Let’s Be Realistic: Crafting an Effective Legal Remedy for Victims of Deepfake Pornography. Ariz. L. Rev. 66 (2024), 785

    2024

  62. [70]

    Gerald R Salancik and Jeffrey Pfeffer. 1974. The bases and use of power in organizational decision making: The case of a university. Administrative science quarterly (1974), 453–473

    1974

  63. [71]

    Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson, and Yarin Gal. 2024. AI models collapse when trained on recursively generated data. Nature 631, 8022 (2024), 755–759

    2024

  64. [72]

    Jason M Solomon. 2011. Civil Recourse as Social Equality. Fla. St. UL Rev. 39 (2011), 243

    2011

  65. [73]

    Arthur L Stinchcombe. 2013. Social structure and organizations1. In Handbook of organizations (RLE: Organizations) . Routledge, 142–193

    2013

  66. [74]

    Mark C Suchman. 1995. Managing legitimacy: Strategic and institutional approaches. Academy of management review 20, 3 (1995), 571–610

    1995

  67. [75]

    Harini Suresh, Emily Tseng, Meg Young, Mary Gray, Emma Pierson, and Karen Levy. 2024. Participation in the age of foundation models. In The 2024 ACM Conference on Fairness, Accountability, and Transparency . 1609– 1621

    2024

  68. [76]

    Thompson

    James D. Thompson. 2017. Organizations in Action: Social Science Bases of Administrative Theory. Routledge, New York. Original work published 1967

    2017

  69. [77]

    Lynne Tirrell. 2013. Apologizing for Atrocity: Rwanda and Recognition . Springer Netherlands, Dordrecht, 159–182. doi:10.1007/978-94-007-5201- 6_10

    2013 doi

  70. [78]

    Rebecca Umbach, Nicola Henry, Gemma Faye Beard, and Colleen M Berryessa. 2024. Non-Consensual Synthetic Intimate Imagery: Preva- lence, Attitudes, and Knowledge in 10 Countries. In Proceedings of the CHI Conference on Human Factors in Computing Systems . 1–20

    2024

  71. [79]

    Sohini Upadhyay, Shalmali Joshi, and Himabindu Lakkaraju. 2021. To- wards robust and reliable algorithmic recourse. Advances in Neural Infor- mation Processing Systems 34 (2021), 16926–16937

    2021

  72. [80]

    Census Bureau

    U.S. Census Bureau. 2023. The Older Population in the United States: Percentage Aged 65 and Over. https://www.census.gov/. Accessed: 2025- 01-22

    2023

  73. [81]

    Ana Valdivia. 2024. The Supply Chain Capitalism of AI: A Call to (Re) think Algorithmic Harms and Resistance. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society , Vol. 7. 1466–1466

    2024

  74. [82]

    Sandra Wachter, Brent Mittelstadt, and Luciano Floridi. 2017. Why a right to explanation of automated decision-making does not exist in the general data protection regulation. International data privacy law 7, 2 (2017), 76–99

    2017

  75. [83]

    Max Weber. 1978. Economy and society: An outline of interpretive sociology . Vol. 1. University of California press

    1978

  76. [84]

    David Gray Widder and Dawn Nafus. 2023. Dislocated accountabilities in the AI supply chain. Big Data & Society 10, 1 (2023). doi:10.1177/ 20539517231162565

    2023

  77. [85]

    Oliver E Williamson. 1985. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. University of Illinois at Urbana- Champaign’s Academy for Entrepreneurial Leadership Historical Research Reference in Entrepreneurship (1985)

    1985

  78. [86]

    Dennis H. Wrong. 1979. Power: Its Forms, Bases, and Uses . Harper and Row, New York

    1979

  79. [87]

    Zikai Xie, Xenophon Evangelopoulos, Ömer H Omar, Alessandro Troisi, Andrew I Cooper, and Linjiang Chen. 2024. Fine-tuning GPT-3 for ma- chine learning electronic and functional properties of organic molecules. Chemical science 15, 2 (2024), 500–510

    2024

  80. [88]

    Jaffer Zaidi. 2025. Working with The Associated Press to provide fresh results for the Gemini app. https://blog.google/products/news/associated- press-gemini-app/ Hopkins et al. Table 2: How each market dynamic affects the achievability and relational power around every route ...

    2025

  81. [1200]

    doi:10.1145/3593013.3594073

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.