<p>Testimonial justice is the virtue guiding our assignment of credibility to a speaker. In the debate on testimonial injustice, it is often understood primarily as a tool to prevent discrimination and generally the unfounded discounting of the credibility of certain speakers and groups. But the function of the virtue extends far beyond the purpose of countering prejudice: It also serves as a safeguard against dishonest agents of various kinds. Such agents threaten the integrity of socio-epistemic processes, including the institutions of science and their communication with policymakers and the general public. We offer a classification of dishonest agents and translate them into a simple agent-based model inhabited by naive Bayesian agents. In this formal framework, we investigate the differential impact of varieties of dishonesty and the utility of testimonial justice in preventing or mitigating distortions as well as reliably identifying dishonest agents, given mildly benign conditions.</p>