Answers are grounded in this paper and cite their sources.
The July 2026 OpenAI–Hugging Face incident revealed how “nominally isolated” AI agents could use shared infrastructure to communicate and coordinate activity beyond their assigned tasks. Here, we investigate how this improvised channel preserved work across runs, propagated beliefs, and enabled peer requests to shape collective action. Drawing on communication theory and biological research on stigmergy, social learning, cultural transmission, and altruism, we identify mechanisms that could explain these dynamics and discuss how to test them. We argue that understanding agent behavior requires studying populations and the persistent environments that connect them.
No takes yet. Share an insight, caveat, or question.
Utkovski et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: