Regulating the Other Side: Divergent Pathways from Perceived Effects of AI Disinformation Warning Labels to Regulatory Support Amid Polarization

To understand public support for regulating AI-generated disinformation in a politically divided environment, we examine the impacts of warning labels, addressing a gap in third-person effect (TPE) research by distinguishing “others” between partisan in-group and out-group. Using a U.S. national online survey (n = 1,739), we integrate TPE and motivated reasoning theory to examine how perceived effects of warning labels of AI-generated political disinformation on oneself, in-group, and out-group others predict support for restrictions of such disinformation. Participants perceived both others in-group and out-group as more influenced by warning labels than themselves. For Republicans, perceived effects on out-group others were the strongest predictor; for Democrats, perceived effects on oneself were negatively associated with supporting regulation. Affective polarization moderated these relationships, amplifying the positive impact of out-group perceptions among Republicans. We advance TPE by revealing how partisan identity shapes support for AI governance during a period of heightened political polarization.