Moral Disengagement

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI addresses the AI transition primarily from the perspective of the worker navigating it. Moral disengagement supplies the complementary perspective: the psychological mechanisms operating in the people making the decisions that workers must navigate. The cycle’s account of the Trivandrum workshop — the engineers who became less tolerant of pre-AI work rhythms within days of experiencing AI-augmented work — is also an account of the conditions that make moral disengagement available to their managers: genuine productivity gains, genuine competitive pressure, genuine belief in the technology’s potential, and the organizational distance between decision-maker and displaced worker that allows the consequences to remain abstract.

The attribution of blame mechanism is particularly visible in the AI transition discourse. The narrative that workers who struggle to adapt are simply “behind the curve,” that displaced workers are “Luddites” who chose resistance over adaptation, converts a structural economic transformation into a personal character failing. This conversion serves multiple psychological functions simultaneously: it disengages the moral responsibility of those who benefit from the transformation, it provides a satisfying causal explanation that requires no institutional response, and it deepens the self-efficacy erosion of the displaced by adding the attribution of personal failure to the structural threat. Bandura’s framework insists that the historical Luddites were not irrational opponents of progress; they were skilled workers who accurately predicted specific harms that the technology would cause to their communities. The dismissal of their concerns as “fear of technology” was itself moral disengagement, and its contemporary equivalent performs the same function.

Origin

Bandura developed the moral disengagement framework across the last decades of his career, building on his broader social cognitive theory to explain how people who possess moral standards can behave in ways that violate those standards without experiencing the self-condemnation that standards are supposed to produce. The key empirical observations that motivated the framework were studies of individuals who participated in atrocities, military violence, organizational fraud, and chronic workplace mistreatment — situations in which people behaved harmfully while maintaining a self-image as moral actors. The framework was formalized in his 2015 book and subsequently applied to organizational contexts, social movements, and the ethics of technology.

The application to the AI transition was not part of Bandura’s own published work, which concluded with his death in 2021 at the age of 95. But his framework’s applicability to the conditions of the AI transition is direct and documented in the book that grounds this entry: the mechanisms he identified operate in precisely the institutional contexts — technology companies, investment decisions, organizational restructuring — where AI deployment decisions are made. The fact that he did not live to apply them himself does not limit their applicability; his framework was designed to be domain-general, and the AI transition provides a domain in which its predictions are directly testable.

Key Ideas

The eight mechanisms. Bandura identified eight distinct cognitive strategies that enable moral disengagement: moral justification, euphemistic labeling, advantageous comparison, displacement of responsibility, diffusion of responsibility, disregard or distortion of consequences, dehumanization, and attribution of blame. Each operates through a different cognitive route. Together they constitute a comprehensive repertoire of moral self-protection. In any given organizational context, multiple mechanisms typically operate simultaneously, and their combination is more effective than any single mechanism would be alone.

Moral justification in the AI transition. The most powerful mechanism, and the most active in the AI transition, is moral justification: the reframing of harmful conduct as serving a higher purpose. The executive who deploys AI tools that eliminate positions does not say “I am choosing to harm these people to increase profits.” She says “This transition is inevitable, and by leading it responsibly, we are positioning the company for long-term success.” The justification contains genuine truth — the transition is real, the competitive pressure is real, the technology’s potential is real — and the truth is precisely what makes the justification effective. Bandura’s framework insists that the truth of the justification does not eliminate the harm, and that genuine moral engagement requires holding both realities simultaneously rather than using the justification to render the harm invisible.

Attribution of blame and the Luddite narrative. Attribution of blame is the moral disengagement mechanism most visible in the AI transition discourse. It converts a structural economic transformation — in which specific decisions made by specific people produce specific harms to specific workers — into a personal failing of the harmed. The displaced worker “should have seen it coming,” “should have adapted faster,” “is clinging to obsolete methods.” Each attribution transfers responsibility from the deployer to the displaced and reinforces the narrative that no institutional response is required. The historical Luddites provide the template: they accurately predicted the specific harms that power looms would cause to their communities, and their concerns were dismissed as irrational resistance to inevitable progress — a dismissal that served the interests of the people benefiting from the technology.

Albert Bandura

Moral engagement as the antidote. Bandura’s prescription is what he calls moral engagement: the active, ongoing confrontation with the human consequences of one’s decisions. Moral engagement requires several practices that the AI deployment culture has not adequately adopted: making the human consequences visible rather than abstract, accepting personal rather than institutional responsibility, and what Bandura called anticipatory self-regulation — foreseeing the consequences of one’s actions and adjusting before the harm occurs rather than after. The prescription is organizational as well as individual: organizations that build moral engagement into their decision processes — requiring decision-makers to confront the human consequences of their choices, rewarding moral engagement rather than normalizing disengagement — produce individuals who maintain moral agency even under pressures that would otherwise produce disengagement.

Self-efficacy for moral action. Bandura connected the moral disengagement framework to his broader self-efficacy theory through a final concept: self-efficacy for moral action, the belief that one has the capacity to maintain moral standards even under pressure to disengage. Like all self-efficacy, this belief is built through mastery experience — the experience of having resisted pressure, spoken up, insisted on moral engagement when the organizational culture pushed toward evasion. Organizations that build this form of self-efficacy, through training, through incentive structures, through creating spaces where moral concerns can be raised without career risk, are building the psychological infrastructure that prevents the default toward moral disengagement that Bandura documented.