Explaining the Reputational Risks of AI-Mediated Communication: Messages labeled as AI-assisted are viewed as less diagnostic of the sender's moral character

AI labels reduce how much we learn about someone's character from their messages. Rather than making people view AI-assisted messages negatively overall, the research found that AI labels simply weaken the connection between what someone writes and what we infer about who they are as a person. This means an AI-assisted apology tells us less about whether someone is genuinely warm and caring.

The same AI label can paradoxically make both positive and negative messages seem less authentic. When someone writes a warm message like an apology without AI help, people judge them as more genuinely warm. But when that same apology includes an AI-assistance label, the sender appears less warm. Conversely, AI labels make harsh messages (like blame) seem less indicative of actual coldness—suggesting the AI label creates ambiguity about the sender's true intentions in both directions.

Transparency about AI use may have unintended reputational consequences. The findings suggest that disclosing AI assistance, while ethically important, comes with a cost: it undermines the ability of sincere communication to demonstrate moral character. Organizations and individuals should consider how transparency requirements might affect how their genuine efforts to communicate thoughtfully are perceived.

Context matters for how AI-assisted communication is interpreted. The research demonstrates that the type of message (warm vs. cold) interacts with AI labels in important ways, suggesting that blanket policies about AI disclosure may not account for how different communication contexts affect reputation and trust.