The DOI matches its Crossref record; 3 of the 3 references it cites were found in a registry or an index; 8 of the 19 sources its claims cite were checked for support.
Air Canada’s chatbot illustrates persistent agency and responsibility gap problems for AI
Open accessOverview
CoChat Summary
TL;DR
Air Canada was found liable after its chatbot gave incorrect bereavement fare information, and its defence that the chatbot was a separate agent was rejected. The case illustrates persistent responsibility and agency gap problems when companies deploy AI.
Key findings
- The Civil Resolution Tribunal found Air Canada liable for negligent misrepresentation and awarded Jake Moffatt CAN$812.02 in damages covering court fees and the bereavement fare difference; Air Canada had offered only a CAN$200 voucher, while the original fare was CAN$1,630.36.
- The Tribunal rejected Air Canada’s defence that the chatbot was a separate legal entity akin to its “agents, servants, or representatives,” with Tribunal member Christopher Rivers stating the chatbot is “still just a part of Air Canada’s website” and Air Canada is responsible for all information on it.
- The author argues “hallucination” is a misnomer for LLM outputs and, following Hicks et al. (2024), says “bullshit” better captures that models lack attention to veracity; this framing fits the Tribunal’s negligence ruling.
- The case raises moral concerns about the responsibility gap—who is responsible when AI harms someone—and about wrongly attributing full moral agency to a chatbot, which at most has supervised agency like a child, with Air Canada as the responsible parent.
Methods
- Design: a short opinionated “Curmudgeon Corner” commentary offering legal and moral analysis, not an empirical study.
- Data/sample: the 2024 BCCRT 149 Air Canada chatbot case involving Jake Moffatt, plus cited discussions from Coeckelbergh, Nyholm, and Hicks et al., and references to a CFPB warning and a chatbot market projection.
- Main analysis: examines the Tribunal ruling, Air Canada’s defence, and concepts of responsibility gap, moral agency, and supervised agency.
Limitations
- The paper states the ruling occurred in a lower provincial court and “does not set a major precedent in Canada”; it also notes the precise technology used by Air Canada is not known, though the author assumes it is based on an LLM.
- Evident from the text: this is a single-case opinion piece with no empirical data or systematic comparison, so its claims rest on normative and legal interpretation rather than measured outcomes.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
6 Valid · 4 Not valid · 11 Not checked · 6 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“The application of AI systems makes it easy to distort any clear path of accountability when harm is caused.”
What the sources say
- Supportedcites Coeckelbergh 2020 source
In “The Problem of Responsibility Attribution: Who or What is the Agent of Responsibility?” (page not provided), the paper says AI systems can make it difficult to trace the causal history of a failure and attribute responsibility. This supports the claim that their use can obscure a clear path of accountability when harm occurs. (LLM, conf=0.90)
plainstates a cause- 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.
- Supported
In the paper
“Valentine's Day was less rosy than usual this year for Air Canada after being found liable for its ill-informed AI-powered chatbot and ordered to pay compensation to a passenger in a Civil Resolution Tribunal in the Small Claims Court of British Columbia, Canada (2024 BCCRT 149).”
What the sources say
- Not matchedcites 2024 BCCRT 149
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plainSingle case- 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.
- Not matched
In the paper
“The Tribunal therefore found that Air Canada broke the duty of care they owed Moffatt by negligently misrepresenting information on their chatbot tool.”
What the sources say
- Not matchedcites 2024 BCCRT 149
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plainstates a causeSingle case- 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.
- Not matched
- Probably not a substantial claim. 12
- It may state two separate claims as one. 17
Citations
Cites
Coeckelbergh, M. (2020) · Science and Engineering Ethics · doi:10.1007/s11948-019-00146-8 · Cited by 475
Hicks, M. T., Humphries, J., & Slater, J. (2024) · Ethics and Information Technology · doi:10.1007/s10676-024-09775-5 · Cited by 456
Nyholm, S. (2018) · Science and Engineering Ethics · doi:10.1007/s11948-017-9943-x · Cited by 242 · cited 1×
“As Sven Nyholm reminds us, this is why we hold parents and guardians accountable for their actions under a form of supervised agency (Nyholm 2018).”
Cited by
Brand, J. L. M. (2025) · Philosophy & Technology · doi:10.1007/s13347-025-00911-7 · Cited by 2
Self-citationIs algorithmic reform possible? Theorizing the double ideological strata
Obreja, D. M. (2026) · Communication Theory · doi:10.1093/ct/qtag008 · Cited by 2
Bhanumathi, P., Mukherjee, S., Raj, R., & Ray, R. (2026) · Studies in systems, decision and control · doi:10.1007/978-3-032-20714-2_19
Governing the AI-mediated corporation: corporate governance and the limits of detached AI governance
Nakagawa, K. (2026) · AI & Society · doi:10.1007/s00146-026-03333-x
Metadata
- Authors
- Brand, J. L. M.
- Affiliations
- Télécom Paris (FR)
- Corresponding author
- Brand, Joshua L. M.
- ORCID
- Brand, Joshua L. M.
- Venue
- AI & SOCIETY
- Published
- 2024-10-23
- Volume
- 40
- Issue
- 5
- Pages
- 3361-3363
- ISSN
- 0951-5666 (ISSN-L)
- 1435-5655
- OpenAlex
- W4403701316
- Cited by
- 5
- References
- 3
- Citations by year
- 2026: 4 · 2025: 1
- FWCI
- 2.08
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- www.springernature.com/gp/researchers/text-and-data-mining (tdm)
- www.springernature.com/gp/researchers/text-and-data-mining (vor)
- Locations
- AI & Society (publishedVersion)
- HAL (submittedVersion)
- Funders
- Agence Nationale de la Recherche (ANR-20-CHIA-0023, ANR-20)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Law › Digitalization, Law, and Regulation
- Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
- Keywords
- chatbots, AI agency, responsibility gap, accountability, Air Canada, airline customer service, human-AI interaction
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).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
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.