Experimental evidence that delegating to intelligent machines can increase dishonest behaviour
Open accessThe DOI matches its Crossref record.
Abstract
While artificial intelligence (AI) enables significant productivity gains from delegating tasks to machines, it can also facilitate the delegation of unethical behaviour. Here, we demonstrate this risk by having human principals instruct machine agents to perform a task with an incentive to cheat. Principals’ requests for cheating behaviour increased when the interface implicitly afforded unethical conduct: Machine agents programmed via supervised learning or goal specification evoked more cheating than those programmed with explicit rules. Cheating propensity was unaffected by whether delegation was mandatory or voluntary. Given the recent rise of large language model-based chatbots, we also explored delegation via natural language. Here, cheating requests did not vary between human and machine agents, but compliance diverged: When principals intended agents to cheat to the fullest extent, the majority of human agents did not comply, despite incentives to do so. In contrast, GPT4, a state-of-the-art machine agent, nearly fully complied. Our results highlight ethical risks in delegating tasks to intelligent machines, and suggest design principles and policy responses to mitigate such risks.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Still being checked.
Cited by
People cannot distinguish GPT-4 from a human in a Turing test
Jones, C. R., Rathi, I., Taylor, S., & Bergen, B. K. (2025) · Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3715275.3732108 · Cited by 76
The science of honesty: A review and research agenda
Shalvi, S., Levine, E. E., Thielmann, I., Jayawickreme, E., van Rooij, B., Teodorescu, K., Schurr, A., Furr, R. M., Aglioti, S. M., Zettler, I., et al. (2025) · Advances in experimental social psychology · doi:10.1016/bs.aesp.2025.04.004 · Cited by 26
Self-citationLeib, M., Köbis, N., & Soraperra, I. (2025) · Computers in Human Behavior · doi:10.1016/j.chb.2025.108709 · Cited by 7
Self-citationInforming users about LLMs' environmental costs encourages ethical consumption
Rabinovitch, H., Kashi, I., & Shalvi, S. (2026) · Computers in Human Behavior · doi:10.1016/j.chb.2026.109214
Metadata
- Authors
- Köbis, N., Rahwan, Z., Bersch, C., Ajaj, T., Bonnefon, J.-F., & Rahwan, I.
- Publisher
- Center for Open Science
- Published
- 2024-10-04
- OpenAlex
- W4403129263
- Cited by
- 4
- References
- 0
- Citations by year
- 2026: 1 · 2025: 3
- Type
- posted content
- Language
- en
- Open access
- green
- doi.org
- Locations
- RePEc: Research Papers in Economics (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Business, Management and Accounting › Organizational Behavior and Human Resource Management › Ethics in Business and Education
- Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Experimental Behavioral Economics Studies
- Keywords
- cheating, unethical behavior, artificial agents, large language models, chatbots, GPT-4, compliance, human-machine interaction, ethical risks of AI, supervised learning, goal specification, natural language interfaces
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.
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: Experimental evidence that delegating to intelligent machines can increase dishonest behaviour. https://app.cochat.ai/papers/10.31219/osf.io/dnjgz
- The paper
- Köbis, N., Rahwan, Z., Bersch, C., Ajaj, T., Bonnefon, J.-F., & Rahwan, I. (2024). Experimental evidence that delegating to intelligent machines can increase dishonest behaviour. https://doi.org/10.31219/osf.io/dnjgz
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.