Theoretical modeling reveals belief attractors propagate through conversational AI interactions, highlighting the need for epistemic defensive architectures.
We examine the claim that 'mind viruses' generated by large language models can become biological pathogens. The primitive assertion fails: a digital information pattern cannot transmute into a biological agent. The derivative phenomenon is real and operationally important: language-mediated belief/behavior attractors can propagate through human-agent conversation, reinforce themselves via model sycophancy and anthropomorphic framing, and produce downstream harm. We model this mechanism as a contagion of attractors and describe a defensive architecture (EPHEMERIS/Warden) with a shared epistemic floor (ladder corpus). We also introduce an Abstract Language Model with a Beer-Lambert defuzzification gate as a novel abstraction firewall.
No takes yet. Share an insight, caveat, or question.
jeffery Harris (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: