Shoppers dealing with AI service agents say they'd be more willing to act unethically

The researchers trace the gap to reduced perceived social judgment: how far consumers think the agent can morally evaluate their behavior. They tested and ruled out entitlement, anticipatory guilt and moral disengagement as explanations.

The effect was strongest when internal accountability (moral identity) and external accountability (detectability) were both low. Priming people to think about being evaluated did not restore the sense of judgment in AI interactions, which the authors read as a perception tied to the agent itself rather than to what is on a person's mind at the moment.

The study asks whether AI-mediated service encounters raise unethical consumer behavior and what psychological mechanism lies behind it. Across a pilot and three experiments, the outcome measured was consumers' reported unethical behavioral intentions.

The mechanism was reduced perceived social judgment, defined as how capable consumers think the interacting agent is of morally evaluating them. Alternative accounts based on entitlement, anticipatory guilt and moral disengagement did not account for the difference.

Accountability shaped the size of the effect. It was strongest where consumers had low moral identity and low detectability. A priming manipulation that raised evaluative concern did not close the gap, so the authors conclude the reduced judgment stems from the agent's perceived evaluative capacity.

The authors say AI changes the evaluative dynamics of retail and service encounters and increases the risk of unethical consumer behavior. They call on firms to embed accountability mechanisms when putting AI in customer-facing roles.