The DOI matches its Crossref record, with a small difference in title or year; 43 of the 43 references it cites were found in a registry or an index; none of the 17 sources its claims cite is checked for support yet.
The poverty of ethical AI: impact sourcing and AI supply chains
Open accessAbstract
Abstract Impact sourcing is the practice of employing socio-economically disadvantaged individuals at business process outsourcing centres to reduce poverty and create secure jobs. One of the pioneers of impact sourcing is Sama, a training-data company that focuses on annotating data for artificial intelligence (AI) systems and claims to support an ethical AI supply chain through its business operations. Drawing on fieldwork undertaken at three of Sama’s East African delivery centres in Kenya and Uganda and follow-up online interviews, this article interrogates Sama’s claims regarding the benefits of its impact sourcing model. Our analysis reveals alarming accounts of low wages, insecure work, a tightly disciplined labour management process, gender-based exploitation and harassment and a system designed to extract value from low-paid workers to produce profits for investors. We argue that competitive market-based dynamics generate a powerful force that pushes such companies towards limiting the actual social impact of their business model in favour of ensuring higher profit margins. This force can be resisted, but only through countervailing measures such as pressure from organised workers, civil society, or regulation. These findings have broad implications related to working conditions for low-wage data annotators across the sector and cast doubt on the ethical nature of AI products that rely on this form of AI data work.
Overview
CoChat Summary
## TL;DR
This paper uses fieldwork at three Sama delivery centres in Kenya and Uganda to test Sama’s claims that its “impact sourcing” model supports ethical AI and good work. It finds that while Sama does create some real economic benefits, workers also faced low pay, insecure contracts, heavy surveillance, and gender-based exploitation; after the fieldwork, Sama reported major improvements and its Fairwork score rose from 0/10 to 5/10.
## Key findings
- The study reports that Sama had impacted 65,343 people by Q1 2023, including 15,083 employees, 41,157 dependents, and 9,312 people who received short digital literacy training.
- At the time of interviews, not all workers were paid Sama’s own living-wage benchmark: in Kenya, one worker showed a payslip for 23,805 KES ($173.28), another earned 28,000 KES ($204.76), and newcomers started on 20,000 KES ($146.26); in Uganda, reported base incomes ranged from 620,000 UGX ($167.40) to 800,000 UGX ($215.20), all below the recommended living wage.
- Working conditions were described as poor and insecure: workers reported 1- or 3-month contracts, some as short as 1 week, 10-hour days, strict performance targets, digital monitoring, unpaid overtime, and occasional 5-to-6-hour unpaid Saturday shifts.
- The paper reports widespread gender-based exploitation, including contract non-renewal for pregnant workers to avoid 3 months of new parent leave, and sexual harassment/pressure from managers in exchange for jobs or security.
## Methods
- Qualitative case study of Sama, based on fieldwork in April and May 2023 at three East African delivery centres in Nairobi, Kenya and Gulu, Uganda, plus follow-up interviews.
- Data sources included workplace observation, facility tours, company presentations/documents, and 46 semi-structured interviews with workers and management; the paper also notes a smaller external sample of current and former Sama employees.
- Interviews were recorded, transcribed, anonymized, and analyzed using inductive thematic analysis focused on Sama’s claims about impact sourcing and worker benefits.
## Limitations
- The paper states that participants did not give written consent for their data to be shared publicly, so supporting data are not available.
- The interview sample was arranged by the company for the on-site interviews, which the text itself suggests may limit independence and representativeness.
- Evident from the text: the study is a single-company qualitative case study of three centres, so the findings are not designed to be statistically generalizable beyond Sama or the broader sector.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 0 Not valid · 14 Not checked · 113 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our analysis reveals alarming accounts of low wages, insecure work, a tightly disciplined labour management process, gender-based exploitation and harassment and a system designed to extract value from low-paid workers to produce profits for investors.”
- Also stated in: § 1 Introduction · plain
“Our analysis from our initial fieldwork revealed alarming accounts of low wages, insecure work, a tightly disciplined labour management process, genderbased exploitation and harassment, and a system designed to extract value from low-paid workers to produce profits for venture capital investors.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Also stated in: § 1 Introduction · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our analysis reveals alarming accounts of low wages, insecure work, a tightly disciplined labour management process, gender-based exploitation and harassment and a system designed to extract value from low-paid workers to produce profits for investors.”
- Also stated in: § 1 Introduction · plain
“Our analysis from our initial fieldwork revealed alarming accounts of low wages, insecure work, a tightly disciplined labour management process, genderbased exploitation and harassment, and a system designed to extract value from low-paid workers to produce profits for venture capital investors.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Also stated in: § 1 Introduction · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our analysis reveals alarming accounts of low wages, insecure work, a tightly disciplined labour management process, gender-based exploitation and harassment and a system designed to extract value from low-paid workers to produce profits for investors.”
- Also stated in: § 1 Introduction · plain
“Our analysis from our initial fieldwork revealed alarming accounts of low wages, insecure work, a tightly disciplined labour management process, genderbased exploitation and harassment, and a system designed to extract value from low-paid workers to produce profits for venture capital investors.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Also stated in: § 1 Introduction · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our analysis reveals alarming accounts of low wages, insecure work, a tightly disciplined labour management process, gender-based exploitation and harassment and a system designed to extract value from low-paid workers to produce profits for investors.”
- Also stated in: § 1 Introduction · plain
“Our analysis from our initial fieldwork revealed alarming accounts of low wages, insecure work, a tightly disciplined labour management process, genderbased exploitation and harassment, and a system designed to extract value from low-paid workers to produce profits for venture capital investors.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Also stated in: § 1 Introduction · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Almost all workers we interviewed reported feeling insecure and fearful about the prospects of losing their job.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
67 more claims
Create a free account to see all 127 claims and every source verdict.
- Probably not a substantial claim. 27
- It may state two separate claims as one. 34
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“AI colonialism thus represents a new form of digital and environmental exploitation, where the benefits accrue to the Global North's super-rich class while the Global South remarkably suffers the externalities.10,11,12,13,14,15,16”
From Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South (2024) · cites this paper as [16]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Companies such as Sama and CloudFactory typically adopt the “impact sourcing” model to cultivate moral economies of “ethical AI” that appeal to clients while imposing precarious labor conditions onto workers [21, 95].”
From 'The plan is just survival': Data Work in Kenya and the Regime of Entrapment (2026) · cites this paper as [95]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“The intensification of exploitation or superexploitation [126] made visible through the regime of entrapment, is evocative; however, there has been exceptional scholarship and journalism documenting the exploitation data workers endure [8, 94, 95, 109, 142].”
From 'The plan is just survival': Data Work in Kenya and the Regime of Entrapment (2026) · cites this paper as [95]
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“Amazon, Google, and Microsoft have constructed data centers in Africa and Southeast Asia, attracted by cheap electricity and limited governance oversight.”
From Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South (2024) · cites this paper as [32]
Experimental — accuracy not yet measuredThe check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“In Kenya, where droughts and environmental crises have worsened recently, energy-intensive data centers by global tech companies raise significant concerns about the long-term sustainability of AI development in these regions.32,33,34,35”
From Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South (2024) · cites this paper as [32]
Experimental — accuracy not yet measured
Citations
Cites
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988 · cited 1×
“…increasingly popular with many companies, there are currently few regulatory requirements for AI companies to disclose any information concerning the provenance of their datasets or other critical infrastructure related to AI systems (Bender et al. 2021; Gebru et al. 2021; Marshall et al. 2015).”
The global landscape of AI ethics guidelines
Jobin, A., Ienca, M., & Vayena, E. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0088-2 · Cited by 6,220 · cited 1×
“Although much research has focussed on ethical issues related to the outcomes of AI systems (Floridi 2019; Hagendorff 2020; Jobin et al. 2019), another key consideration is the ethics of how AI systems are produced.”
The SAGE Encyclopedia of Qualitative Research Methods
Given, L. (2008) · doi:10.4135/9781412963909 · Cited by 6,134
Gebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., Daumé, H., & Crawford, K. (2018) · arXiv · doi:10.48550/arxiv.1803.09010 · Cited by 2,474
Gebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., III, H. D., & Crawford, K. (2021) · Communications of the ACM · doi:10.1145/3458723 · Cited by 2,469
Superintelligence: paths, dangers, strategies
(2015) · Choice Reviews Online · doi:10.5860/choice.187536 · Cited by 1,979
The Ethics of AI Ethics: An Evaluation of Guidelines
Hagendorff, T. (2020) · Minds and Machines · doi:10.1007/s11023-020-09517-8 · Cited by 1,911 · cited 1×
“Although much research has focussed on ethical issues related to the outcomes of AI systems (Floridi 2019; Hagendorff 2020; Jobin et al. 2019), another key consideration is the ethics of how AI systems are produced.”
The Ethics of AI Ethics -- An Evaluation of Guidelines
Hagendorff, T. (2019) · arXiv · doi:10.48550/arxiv.1903.03425 · Cited by 1,907 · cited 1×
“Although much research has focussed on ethical issues related to the outcomes of AI systems (Floridi 2019; Hagendorff 2020; Jobin et al. 2019), another key consideration is the ethics of how AI systems are produced.”
Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence
Crawford, K. (2021) · doi:10.12987/9780300252392 · Cited by 1,361
(2019) · Cited by 1,082
Research works printed but not linked: 8
- Cloudfactory (2023) In Their own words: an interview with CloudFactory. In: In Their Own Words: An Interview with CloudFactory. Available at: https://www.cloudfactory.com/hs-fs/hub/351374/file-1151354869-pdf/press-files/gscouncil-In_Their_Own_Words_An_Interview_with_CloudFactory.pdf (Accessed 1 Dec 2023)
- Graham M, Woodcock J (2018) Towards a fairer platform economy: introducing the fairwork foundation. Alternate Routes 29:242–253
- Janah L (2017) Give work: reversing poverty one job at a time. New York: Penguin Random House. Available at: https://www.penguinrandomhouse.com/books/546305/give-work-by-leila-janah/ (Accessed 9 May 2023)
- Prahalad C (2004) The fortune at the base of the pyramid. Wharton School Publishing, Upper Saddle River
- Schmidt F (2019) Crowdproduktion von trainingsdaten: Zur Rolle von Online-Arbeit beim Trainieren autonomer Fahrzeuge. Berlin: Hans-Böckler Stiftung. Available at: https://www.researchgate.net/publication/332964049
- Varsha (2019) Samasource Secures $14.8M Series A Funding to Further Scale Business and AI Capabilities. In: SAMA. Available at: https://www.sama.com/press/samasource-secures-14-8m-series-a-funding-to-further-scale-business-and-ai-capabilities/ (Accessed 11 May 2023)
- Woodcock J, Graham M (2019) The gig economy: a critical introduction. Polity Press, Cambridge
- Silverman D (2010) Doing qualitative research: a practical handbook. 3rd ed., London: SAGE publications.
Cited by
Regilme, S. S. F. (2024) · SAIS Review of International Affairs · doi:10.1353/sais.2024.a950958 · Cited by 65
“10,11,12,13,14,15,16 Neoliberal capitalism drives technological innovation for profit and efficiency, often neglecting labor rights, environmental protections, and socioeconomic justice.”
Valdivia, A. (2024) · Information Communication & Society · doi:10.1080/1369118x.2024.2420021 · Cited by 109
“…chains make visible the proliferation of workers in different industrial sectors that might belong to different class and might be exposed to different mechanisms of exploitation and violence (Miceli & Posada, 2022; Muldoon et al., 2023; Posada, 2021; Precarity Lab, 2020; Wang & Jiang, 2021).”
A typology of artificial intelligence data work
Muldoon, J., Cant, C., Wu, B., & Graham, M. (2024) · Big Data & Society · doi:10.1177/20539517241232632 · Cited by 50
“This project involved fi eldwork conducted in 2023 with a BPO specialising in AI data work in Kenya and Uganda (Muldoon et al., 2023).”
InfluentialSelf-citationWorker-led AI governance: Hollywood writers' strikes and the worker power
Grohmann, R., Rocha, A. C., & Guilherme, G. (2025) · Information Communication & Society · doi:10.1080/1369118x.2025.2521375 · Cited by 35
Artificial Intelligence (AI) and Global Justice
Sahebi, S., & Formosa, P. (2024) · Minds and Machines · doi:10.1007/s11023-024-09708-7 · Cited by 27
“…by low-income data workers in the AI pipeline, as well as the lack of mental health support provided by corporations who benefit from their work (Perrigo, 2023 ; Muldoon et al., 2023 ), also puts downward pressure on the ability of people to achieve the threshold to fulfil these Capabilities.”
Influential
Metadata
- Authors
- Muldoon, J., Cant, C., Graham, M., & Ustek Spilda, F.
- Affiliations
- University of Essex (GB)
- University of Oxford (GB)
- Corresponding author
- Muldoon, James
- Venue
- AI & SOCIETY
- Published
- 2023-12-20
- Volume
- 40
- Issue
- 2
- Pages
- 529-543
- ISSN
- 0951-5666 (ISSN-L)
- 1435-5655
- OpenAlex
- W4389991121
- Cited by
- 88
- References
- 38
- Citations by year
- 2026: 42 · 2025: 33 · 2024: 12 · 2023: 1
- FWCI
- 6.27
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- AI & Society (publishedVersion, cc-by)
- Research Portal (King's College London) (submittedVersion, cc-by)
- Open Access at Essex (University of Essex) (submittedVersion)
- Oxford University Research Archive (ORA) (University of Oxford) (submittedVersion, cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Sociology and Political Science › Digital Economy and Work Transformation
- Social Sciences › Business, Management and Accounting › Strategy and Management › Global trade, sustainability, and social impact
- Keywords
- AI supply chain, ethical AI, Sama, data annotation, low-wage employment, business process outsourcing, labor exploitation, gender-based harassment, insecure work, labor management, profit margin, social impact, East Africa, Kenya, Uganda
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: The poverty of ethical AI: impact sourcing and AI supply chains. https://app.cochat.ai/papers/10.1007/s00146-023-01824-9
- The paper
- Muldoon, J., Cant, C., Graham, M., & Ustek Spilda, F. (2025). The poverty of ethical AI: impact sourcing and AI supply chains. AI & SOCIETY. https://doi.org/10.1007/s00146-023-01824-9
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.