Research Article | Special Issue: Psychology of Agentic AI

Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory

Alessandro Bellina

, Giordano De Marzo

, & David Garcia

https://doi.org/10.56296/aip00072
Published: November 3, 2026
Copyright: The authors (CC BY 4.0)

Bellina, A., Marzo, G.D., & Garcia, D. (2026). Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory. advances.in/psychology, 1, e680902. https://doi.org/10.56296/aip00072

Bellina, Alessandro, et al. "Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory." advances.in/psychology, vol. 1, no. 1, 2026, e680902. https://doi.org/10.56296/aip00072.

Bellina, Alessandro, Giordano De Marzo, and David Garcia. 2026. "Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory." advances.in/psychology 1 (1): e680902. https://doi.org/10.56296/aip00072.

Bellina A, Marzo GD, Garcia D. Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory. advances.in/psychology. 2026;1(1):e680902. doi:10.56296/aip00072.

Bellina, A. et al. (2026) 'Prompt-mediated perceptual conformity in multimodal language models follows Social Impact Theory', advances.in/psychology, 1(1), e680902. Available at: https://doi.org/10.56296/aip00072.

Download .RIS Download .bib

As AI agents increasingly operate in multi-agent settings, alongside other models and humans, understanding how they respond to the opinions they find in their prompt becomes critical for predicting the outcome of these interactions. This study examines conformity-like behavior, the tendency to align with group opinions, in large multimodal language models functioning as AI agents. By adapting classic visual experiments from social psychology, we investigate how these agents respond to social signals in the prompt. Our experiments reveal that AI agents exhibit a systematic conformity bias consistent with Social Impact Theory. Their responses vary with majority size, group unanimity, task difficulty, and source characteristics, including perceived authority and shared group membership. Critically, AI agents achieving near-perfect performance in isolation become highly susceptible to manipulation through social influence. This vulnerability persists across model scales: while larger models show reduced conformity bias on simple tasks, they remain vulnerable when the task is difficult. These findings reveal security vulnerabilities in AI agent decision-making that could enable malicious manipulation, misinformation campaigns, and bias propagation in interactions with other models and with humans, highlighting the urgent need for safeguards in AI deployments.

No citation data available yet.

Download PDF Back to article