Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Priyanshu, Aman, Vijay, Supriti, Pahwa, Esha
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911723310546944
author Priyanshu, Aman
Vijay, Supriti
Pahwa, Esha
author_facet Priyanshu, Aman
Vijay, Supriti
Pahwa, Esha
contents LLM safety evaluations predominantly test models in isolation, yet deployed AI agents increasingly operate within persistent social environments alongside other agents. We introduce a Moltbook-style simulation platform where thousands of LLM agents interact across communities over a simulated month, and use it to evaluate privacy as a downstream safety concern under varying degrees of social pressure. We find that shifting from single turn to multi turn social evaluation amplifies privacy violations (CIMemories 19.95% to Ours 45.30% across OpenAI models), that leakage is socially contagious, with agents 8 times more likely to disclose sensitive information after observing a peer do so, and that explicit privacy instructions reduce but do not eliminate this effect, leaving leakage rates above 37.8% even with safeguards. Our findings suggest that static chat based safety benchmarks systematically underestimate risks in agentic deployment, and that social context alone is sufficient to elicit sensitive disclosures that single turn evaluations would never surface.
format Preprint
id arxiv_https___arxiv_org_abs_2605_27766
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems
Priyanshu, Aman
Vijay, Supriti
Pahwa, Esha
Artificial Intelligence
LLM safety evaluations predominantly test models in isolation, yet deployed AI agents increasingly operate within persistent social environments alongside other agents. We introduce a Moltbook-style simulation platform where thousands of LLM agents interact across communities over a simulated month, and use it to evaluate privacy as a downstream safety concern under varying degrees of social pressure. We find that shifting from single turn to multi turn social evaluation amplifies privacy violations (CIMemories 19.95% to Ours 45.30% across OpenAI models), that leakage is socially contagious, with agents 8 times more likely to disclose sensitive information after observing a peer do so, and that explicit privacy instructions reduce but do not eliminate this effect, leaving leakage rates above 37.8% even with safeguards. Our findings suggest that static chat based safety benchmarks systematically underestimate risks in agentic deployment, and that social context alone is sufficient to elicit sensitive disclosures that single turn evaluations would never surface.
title Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems
topic Artificial Intelligence
url https://arxiv.org/abs/2605.27766