Source-linked AI summary
When the Algorithm Becomes the Brand Crisis: A Sociotechnical Theory of Distributed Responsibility and Accountable Transparency
Mohammad Saleh Torkestani, Taha Mansouri
TL;DR
The paper addresses how responsibility is assigned when multiple actors are relevant to an AI-related error rather than treating it solely as a conventional single-actor problem. It develops a framework for studying responsibility networks and finds that AI does not uniformly amplify reputational harm.
Problem
The paper examines responsibility when each of multiple actors may become relevant to an AI-related error rather than treating it solely as a conventional single-actor problem.
Method
The paper introduces a nested construct system for analyzing AI-related incidents and their consequences.
Results
AI does not uniformly amplify reputational harm, and consumer responses can be less negative when an algorithm is involved.
Takeaways & Limitations
The paper shifts analysis from whether “the AI” or “the brand” is blamed toward how responsibility networks shape explanation and evaluation.
Takeaways & Limitations
The paper treats its account of responsibility networks as an empirical prediction rather than an established universal effect.
Abstract
from arXiv · showhide
Artificial intelligence systems increasingly enact market-facing promises through chatbots, recommendation systems, automated decisions, and generative interfaces. Their failures, misuse, and misrepresentation raise a question that conventional brand-crisis models do not fully specify: how do stakeholders assign responsibility when technical causation, customer-facing control, and governance duties are distributed across an AI system, developer, deployer, vendor, and user? This conceptual paper develops a sociotechnical process theory from a structured, federated scoping synthesis of verified academic and primary sources. It distinguishes an AI/algorithmic incident from an AI-related organisational crisis and, in turn, from an AI-related organisational scandal. The framework proposes that incident configuration shapes actor-specific attribution; attribution informs capability, integrity, fairness, and relationship appraisals; and public moralisation may, but need not, escalate an incident into scandal. The theory offers a reconciliation of findings that algorithm involvement can buffer negative brand reactions in some settings while robot and chatbot failures can redirect responsibility to an associated firm in others. It introduces accountable transparency as a proposed response configuration that combines timely notice, an intelligible account, role-responsibility acknowledgement, remedy, evidence of correction, and recourse. The evidence supports conditional, proximal inferences about blame, trust, satisfaction, firm evaluation, and communication credibility more strongly than claims about durable reputation, brand equity, or market performance.
A Sociotechnical Theory of Distributed Responsibility and Transparency
The paper develops a sociotechnical theory that distinguishes AI incidents, organisational crises, and scandals while modelling responsibility across distributed actors. It proposes accountable transparency as a response configuration and bounds its claims primarily to immediate stakeholder appraisals.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: Evidence does not support the conclusion that AI uniformly amplifies reputational harm.Eight experiments found less negative consumer responses to some AI-related harm errors, while robot and chatbot failures could assign more responsibility to an associated firm.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: The framework separates an AI/algorithmic incident from an AI-related organisational crisis and an AI-related organisational scandal.This distinction prevents conflating technical events with crisis demand, public moralisation, and durable reputation loss.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: Responsibility is distributed across actors and includes causal contribution, role responsibility, moral blame, perceived legal duty, and directed responsibility.The framework rejects a binary choice between blaming the AI and blaming a human actor.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: Incident configuration shapes actor-specific attribution, which informs capability, integrity, fairness, and relationship appraisals.An AI output may therefore be reinterpreted as a judgment about organisational character.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: The paper proposes accountable transparency as a future-facing response construct whose evaluation depends on its configuration rather than disclosure alone.The contribution is a specification rather than a claim that AI invariably intensifies scandal.
- A Sociotechnical Theory of Distributed Responsibility and Transparency: The direct literature supports immediate appraisals more strongly than durable reputation, brand equity, sales, employee trust, or market performance.The paper therefore calls for field and longitudinal tests of broader outcomes.
Construct System and Evidence Boundary
The paper proposes nested constructs that distinguish AI-related incidents, organisational crises, and scandals, while separating explanatory objects from possible outcomes. Its evidence boundary supports conceptual, conditional inference rather than exhaustive or pooled empirical claims.
- The construct system distinguishes incidents, crises, and scandals from outcomes such as trust, firm evaluation, legitimacy, and durable brand equity.
- An AI-related organisational scandal crosses an escalation threshold when an episode diffuses, is morally framed, and stimulates institutional accountability demands.
- An organisational crisis can remain private or contained, and public attention is neither a proxy for harm nor evidence of durable damage.
- AI involvement is defined by material relevance rather than probabilism, adaptivity, generativity, embodiment, or autonomy.
- The synthesis was structured and federated but was not a fully systematic review and does not claim exhaustive coverage, pooled effects, or prevalence.
A Responsibility
Responsibility is modelled as distributed across an incident configuration rather than treated as a zero-sum contest between humans and AI. The framework links actor-specific attribution to evaluations, response, and optional scandal escalation.
- A Responsibility: Incident configuration includes harm, risk, severity, stakeholder vulnerability, prior promises, deployment conditions, visibility, recurrence, and control.
- A Responsibility: Control over selection, configuration, interface, monitoring, and redress can cue organisational role responsibility beyond the AI’s causal contribution.
- A Responsibility: Attribution separates causal contribution, role responsibility, blame, perceived legal responsibility, and desired sanction across relevant actors.
- A Responsibility: Evidence describes both buffered brand reactions to algorithm involvement and redirected responsibility to firms after robot or chatbot failures.
- Stage 5: Organisational response and accountable transparency: Accountable transparency combines notice, intelligible explanation, responsibility acknowledgement, remedy, corrective evidence, and accessible recourse.
- Stage 6: Optional scandal escalation and outcome differentiation: Scandal escalation is an optional pathway requiring public diffusion, moralisation, organisational attribution, and institutional accountability demands.
Illustrative Cases and the Limits of Case Inference
The illustrative cases make distributed sociotechnical actor chains visible, but they are used to develop theory rather than establish prevalence, mechanisms, or durable brand effects.
- The cases illustrate possible responsibility patterns but do not prove mechanisms, estimate prevalence, or establish inevitable durable brand damage.
- Primary and authoritative records expose links among customer interfaces, safety programmes, data practices, vendor relationships, and representation.
- The records do not support a uniform vendor shield; deployment responsibility, stakeholder attribution, and available remedy remain analytically important.
Implications for Marketing Theory
The framework extends marketing theory from an isolated-firm view toward relational responsibility across developers, deployers, vendors, users, stakeholders, and perceived AI agents. It also proposes governability and accountable transparency as constructs for future testing.
- From a focal firm to a responsibility network: Actor roles matter because differing capabilities, decision rights, customer access, information, and redress may cue distinct responsibility judgments.
- From a focal firm to a responsibility network: AI-facing promises are enacted through linked relationships among organisations, stakeholders, and associated developers rather than through a focal firm alone.
- Governability and accountable transparency: Governability is proposed as perceived organisational capacity to identify risks, constrain use, document decisions, monitor performance, correct harm, and provide recourse.
- From disclosure to accountable transparency: Disclosure alone is underspecified; the proposed transparency configuration connects intelligible accounts with procedural fairness, corrective action, and stakeholder relationship maintenance.
- From disclosure to accountable transparency: Future studies should test whether accountable-transparency components are formative, substitute for one another, or produce different effects across settings.
Managerial and Policy Implications
The paper recommends governance practices that clarify responsibility, support evidence-based response, and avoid treating compliance as a guarantee of crisis prevention or reputation repair.
- Scope of implications: The framework does not establish a universal reputation effect, and governance activities are not guarantees that a crisis will be prevented or repaired.
- Precautionary governance: Before deployment, organisations can map the actor and control chain, identify foreseeable failure classes, and document monitoring, versioning, and vendor responsibilities.
- Incident response: Managers should distinguish technical causation, organisational role responsibility, and legal fault when responding to incidents.
- Accountable response: Prompt stakeholder acknowledgement, evidence preservation, immediate remedy where possible, and accurate communication support response credibility.
- Legal and contextual boundaries: Disclosure should not prejudge legal liability or final root cause, and applicable law and appropriate disclosure vary by jurisdiction, task, contract, and harm.
- Responsibility allocation: Transferring blame to a vendor or AI system without explaining the organisation’s customer-facing role may create additional integrity concerns.
Limitations and Conclusion
The paper presents a theory whose distinctive claims require direct validation. Its synthesis supports restraint about generalising from heterogeneous, purposively selected evidence, while reframing AI failures as changes in responsibility explanation.
- Limitations: The federated scoping synthesis is not a systematic review because it lacked duplicate screening, formal risk assessment, database denominators, and reproducible numerical flow.
- Limitations: The direct evidence base is heterogeneous, concentrated on service, chatbot, and robot settings, and selected for conceptual variation rather than representativeness.
- Limitations: Regulatory actions, settlements, and company accounts have different evidentiary status and cannot be combined as a uniform measure of stakeholder judgment.
- Conclusion: The framework’s distinctive claims about class foreseeability, multi-actor responsibility, accountable transparency, scandal escalation, governability, and governance require direct validation.
- Conclusion: The central claim is that AI changes the architecture of explanation: stakeholders judge a brand’s control, role, candour, and capacity for remedy alongside the model-connected output.
- Conclusion: An error may remain a contained service failure, become an organisational crisis, or become a public scandal when moralised and amplified across audiences.
- Conclusion: This pathway shifts analysis from whether the AI or brand is blamed toward how responsibility networks shape market relationships under technological uncertainty.
Regis
The passage introduces the paper’s conceptual query architecture without specifying its components.
- The paper identifies a conceptual query architecture.
Role
The paper uses a structured query and coding process to assemble evidence connecting AI systems with organisational, brand, and stakeholder outcomes.
- Search architecture: The conceptual query combines AI and algorithmic-system terms with organisational, brand, crisis, trust, legitimacy, blame, and accountability terms.
- Search architecture: Additional topic blocks pair algorithmic systems with responsibility, controllability, agency, service failure, recovery, disclosure, AI washing, privacy, bias, safety, and crisis.
- Inclusion logic: Sources were retained for direct AI and organisational outcomes, closely related stakeholder mechanisms, or transferable theoretical foundations.
- Case selection: Case records required material AI implication, an identifiable organisation, and an authoritative source documenting the incident, regulatory posture, or related evidence.
- Evidence coding: Evidence was coded as direct, transferred, or new proposition, with direct evidence supporting only outcomes actually measured.
- Inferential boundaries: The synthesis cautions against using a case record or harm result as proof of a long-term or universal effect.