The DOI matches its Crossref record; 14 of the 15 references it cites were found in a registry or an index; 1 of the 23 sources its claims cite were checked for support.
Delegated agency and moral responsibility in artificial intelligence
Open accessAbstract
Introduction: Artificial intelligence ethics is often framed as a response to unprecedented technical autonomy, with risks attributed to recent advances in machine learning and scale. This framing overlooks a recurring ethical structure: the delegation of moral authority to artificial agents. Ethical failures associated with AI are best understood as governance failures rooted in human design choices and accountability arrangements, even where opacity and limited control complicate responsibility attribution. Methods: A qualitative, interdisciplinary approach integrates historical-thematic analysis, comparative interpretation of technological artifacts, and visual-conceptual synthesis. Mythological figures (Talos, the Golem, Pygmalion), early mechanical automata, and foundational computational systems are analyzed as conceptual models of delegated artificial agency rather than technological precursors. Results: Across historical contexts, artificial agents exhibit consistent structural features: bounded autonomy, delegated authority, explicit override mechanisms, and dependence on human oversight. These features directly correspond to contemporary AI ethics concerns, including alignment failures, responsibility gaps, human-in-the-loop control, and system interruptibility. Discussion: The analysis establishes that ethical risk in AI arises from the displacement of human responsibility rather than from machine autonomy. By situating AI within a longer history of artificial agency, the study provides a normative framework that locates moral responsibility unambiguously in human actors and institutions, with direct implications for AI governance and accountability.
Overview
CoChat Summary
## TL;DR
The paper argues that AI ethics problems are best understood as governance failures caused by humans delegating moral authority to artificial systems, not as a new problem created by machine autonomy. Using mythological and historical cases like Talos, the Golem, automata, and Pygmalion, it claims responsibility must remain with human designers, deployers, and institutions.
## Key findings
- Across the cases analyzed, artificial agents repeatedly show the same structural features: bounded autonomy, delegated authority, explicit override mechanisms, and dependence on human oversight.
- The paper argues that alignment problems are governance problems before they are technical ones; misalignment comes from under-specified or conflicting human objectives, not emergent machine morality.
- It states that responsibility gaps arise from distributed human agency—engineers, deployers, executives, regulators, and users—when accountability pathways are unclear.
- The normative model specifies five criteria for proper delegation in frontier systems: auditability, interruptibility, reversibility, capability bounding, and compute monitoring/reporting.
## Methods
- The study uses a qualitative, interdisciplinary, historically grounded interpretive methodology combining historical-thematic analysis, comparative interpretation of technological artifacts, and visual/conceptual synthesis.
- The sample/cases are mythological figures (Talos, the Golem, Pygmalion), mechanical automata (including Al-Jazari and Jaquet-Droz), and early computational systems (Babbage and Lovelace).
- The main analysis compares cases using three criteria: mandate structure, control/override mechanisms, and failure mode, then applies those structural findings to contemporary AI ethics.
## Limitations
- The paper states that it is a normative and conceptual argument rather than a systematic literature review.
- It says the cases are not exhaustive representations of artificial beings in cultural history and were selected for structural clarity over chronological exhaustiveness.
- Evident from the text: the analysis is thematic rather than empirical, so it does not provide original experiments, datasets, or quantitative validation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures























Verification
Claims the paper makes
1 Valid · 14 Not valid · 16 Not checked · 153 Own results and interpretations
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“A historically informed and conceptually grounded approach clarifies that existing ethical principles, such as accountability, oversight, and alignment, do not fail due to conceptual insufficiency, but due to systematic displacement of responsibility.”
§ Delegated agency and moral responsibility in artificial intelligence > Introduction
boostedstates a cause- 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: unclear — It is unclear why the paper says it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These features directly correspond to contemporary AI ethics concerns, including alignment failures, responsibility gaps, human-in-the-loop control, and system interruptibility.”
§ Delegated agency and moral responsibility in artificial intelligence > Abstract > Results
plainstates an association- 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: unclear — It is unclear why the paper says it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Ethical failures associated with AI are best understood as governance failures rooted in human design choices and accountability arrangements, even where opacity and limited control complicate responsibility attribution.”
- Also stated in: § Introduction · plain
“Harmful outcomes attributed to AI are better explained by human design choices, institutional incentives, and governance failures than by machine autonomy itself.”
- Also stated in: § Scholarly contribution and argumentative positioning · plain
“the ethical risks associated with AI stem less from machine autonomy than from human design choices, governance failures, and moral outsourcing.”
§ Delegated agency and moral responsibility in artificial intelligence > Abstract > Introduction
plainstates a cause- 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.
- Also stated in: § Introduction · plain
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The central ethical problem is therefore not whether AI systems possess moral agency, but how moral responsibility is allocated, constrained, and enforced when authority is delegated to artificial systems operating at speed, scale, and organizational distance.”
§ Delegated agency and moral responsibility in artificial intelligence > Introduction
plain- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Delegated moral authority, when combined with scale and abstraction, renders human accountability opaque unless explicitly constrained by institutional mechanisms.”
§ Delegated agency and moral responsibility in artificial intelligence > Introduction
plainstates a cause- 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: unclear — It is unclear why the paper says it holds.
32 more claims
Create a free account to see all 184 claims and every source verdict.
- It may state two separate claims as one. 65
- Probably not a substantial claim. 126
Citations
Cites
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×
“(Floridi et al., 2018; Bartoletti, 2019; Jobin et al., 2019; Milano et al., 2020; IEEE, 2023; Partnership on AI, 2023), the complexity of the ethical landscape and the potential for misuse continue to raise concerns (Mittelstadt, 2019; Roberts et al., 2021; Mökander et al., 2024).”
Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., Luetge, C., Madelin, R., Pagallo, U., Rossi, F., et al. (2018) · Minds and Machines · doi:10.1007/s11023-018-9482-5 · Cited by 4,774 · cited 3×
“Contemporary AI ethics scholarship has addressed alignment (Russell, 2022), responsibility gaps, governance frameworks (Tabassi, 2023a), and value pluralism (Floridi et al., 2018).”
Principles alone cannot guarantee ethical AI
Mittelstadt, B. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0114-4 · Cited by 1,886 · cited 1×
“(Floridi et al., 2018; Bartoletti, 2019; Jobin et al., 2019; Milano et al., 2020; IEEE, 2023; Partnership on AI, 2023), the complexity of the ethical landscape and the potential for misuse continue to raise concerns (Mittelstadt, 2019; Roberts et al., 2021; Mökander et al., 2024).”
Wallach, W., & Allen, C. (2009) · Oxford University Press · doi:10.1093/acprof:oso/9780195374049.001.0001 · Cited by 711 · cited 2×
“Other studies explore the possibility of “machine ethics” through computational implementation of moral decision procedures (Wallach and Allen, 2009).”
The Chinese approach to artificial intelligence: an analysis of policy, ethics, and regulation
Roberts, H., Cowls, J., Morley, J., Taddeo, M., Wang, V., & Floridi, L. (2021) · AI & SOCIETY · doi:10.1007/s00146-020-00992-2 · Cited by 573 · cited 1×
“(Floridi et al., 2018; Bartoletti, 2019; Jobin et al., 2019; Milano et al., 2020; IEEE, 2023; Partnership on AI, 2023), the complexity of the ethical landscape and the potential for misuse continue to raise concerns (Mittelstadt, 2019; Roberts et al., 2021; Mökander et al., 2024).”
Recommender systems and their ethical challenges
Milano, S., Taddeo, M., & Floridi, L. (2020) · AI & SOCIETY · doi:10.1007/s00146-020-00950-y · Cited by 556 · cited 1×
“(Floridi et al., 2018; Bartoletti, 2019; Jobin et al., 2019; Milano et al., 2020; IEEE, 2023; Partnership on AI, 2023), the complexity of the ethical landscape and the potential for misuse continue to raise concerns (Mittelstadt, 2019; Roberts et al., 2021; Mökander et al., 2024).”
The Machine Question: Critical Perspectives on AI, Robots, and Ethics
Gunkel, D. J. (2012) · doi:10.7551/mitpress/8975.001.0001 · Cited by 449 · cited 2×
“…reflect a synthesis of positions in contemporary debates on moral agency and machine ethics (Wallach and Allen, 2009; Coeckelbergh, 2010, 2012; Gunkel, 2012; Parviainen and Coeckelbergh, 2021), while adopting a more restrictive criterion that ties moral agency to intentionality, normative…”
The Machine Question: Ethics, Alterity, and Technology
Gunkel, D. J. (2007) · Series in Philosophy/Communication · doi:10.5840/philcomm200745 · Cited by 440
Robot rights? Towards a social-relational justification of moral consideration
Coeckelbergh, M. (2010) · Ethics and Information Technology · doi:10.1007/s10676-010-9235-5 · Cited by 391 · cited 2×
“These definitions reflect a synthesis of positions in contemporary debates on moral agency and machine ethics (Wallach and Allen, 2009; Coeckelbergh, 2010, 2012; Gunkel, 2012; Parviainen and Coeckelbergh, 2021), while adopting a more restrictive criterion that ties moral agency to intentionality,…”
Auditing large language models: a three-layered approach
Mökander, J., Schuett, J., Kirk, H. R., & Floridi, L. (2024) · AI and Ethics · doi:10.1007/s43681-023-00289-2 · Cited by 289
Research works printed but not linked: 14
- Anderson, 2006, Amazing Leonardo da Vinci Inventions: You Can Build Yourself
- Babbage, 1833, Difference Engines, and Analytical Engines
- Bokser, 1954, From the World of the Cabbalah
- Bruderer, 2019, Leonardo da Vinci’s robot lion, Commun. ACM
- Dodge, 2012, St. Nicholas: A Monthly Magazine for Boys and Girls
- Froelings, 2018, How machine learning works and why it’s important, Payments J.
- Ginsburg, 1997, Clay Boy: Adapted from a Russian Folk Tale
- Hamilton, 1953, Mythology
- Hollings, Ada Lovelace: the making of a computer scientist
- 2023, IEEE introduces new program for free access to AI ethics and governance standards
- Payes, 2019, Leonardo da Vinci’s Lion Robot
- Rodin, 1889, Pygmalion and Galatea. The Met Fifth Avenue in Gallery 800
- Saliba, 2022, Revisiting the Function of Mechanical Devices in Islamic Civilization
- 2014, Mysteries of the Unknown: Inside the World of the Strange and Unexplained
Metadata
- Authors
- Radanliev, P.
- Affiliations
- University of Oxford (GB)
- Corresponding author
- Radanliev, Petar
- ORCID
- Radanliev, Petar
- Venue
- Frontiers in Artificial Intelligence
- Publisher
- Frontiers Media SA
- Published
- 2026-04-15
- Volume
- 9
- Pages
- 1800302-1800302
- ISSN
- 2624-8212 (ISSN-L)
- PMC
- PMC13125054
- OpenAlex
- W7154531124
- Cited by
- 1
- References
- 15
- Citations by year
- 2026: 1
- FWCI
- 5.63
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Frontiers in Artificial Intelligence (publishedVersion, cc-by)
- PubMed (publishedVersion)
- PubMed Central (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Bill and Melinda Gates Foundation (INV-057591)
- Engineering and Physical Sciences Research Council (EP/W020408/1, EP/S035362/1)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
- Social Sciences › Law › Literary and Philosophical Studies
- Keywords
- moral responsibility, AI ethics, AI governance, accountability, human oversight, human-in-the-loop control, machine autonomy, AI alignment, Talos, Golem, Pygmalion, mechanical automata, attribution of responsibility
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
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: Delegated agency and moral responsibility in artificial intelligence. https://app.cochat.ai/papers/10.3389/frai.2026.1800302
- The paper
- Radanliev, P. (2026). Delegated agency and moral responsibility in artificial intelligence. Frontiers in Artificial Intelligence. https://doi.org/10.3389/frai.2026.1800302
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.