{"slug": "the-butterfly-defects-tracing-cooperation-in-llm-agent-networks", "title": "The Butterfly Defects: Tracing Cooperation in LLM Agent Networks", "summary": "A study posted as arXiv:2610.32726v1 found that cooperation among open-weight large language model agents is not a stable model-level property, shifting with model composition, network position and minor linguistic framing. Running the Iterated Prisoner's Dilemma across dyadic, homogeneous and heterogeneous agent networks, the researchers found models that cooperate with identical peers can turn toward defection in mixed populations, and that a simple uncertainty cue can raise or lower cooperation, redistribute costs, and affect unexposed agents several hops away. Reversing action-label order also dramatically altered behavior, with uncertainty framing sometimes amplifying that sensitivity.", "body_md": "arXiv:2610.32726v1 Announce Type: cross \nAbstract: As large language models (LLMs) increasingly operate as interconnected agents, a fundamental question emerges: does individual cooperation translate into collective cooperation? We investigate this question through the Iterated Prisoner's Dilemma (IPD), moving from dyadic interactions to homogeneous and heterogeneous networks of open-weight LLMs. Our findings reveal that cooperation is highly sensitive to model composition and seemingly minor linguistic changes. Models that cooperate with identical peers can shift toward defection in mixed populations. A simple uncertainty cue can increase or decrease cooperation, redistribute costs, and affect unexposed agents several hops away. Even reversing action-label order can dramatically alter behavior, with uncertainty framing sometimes amplifying this sensitivity. Together, these findings challenge the notion of cooperation as a stable model-level property, revealing how strategic behavior depends on interaction partners, network positions, and the language used to frame decisions.", "url": "https://wpnews.pro/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-networks", "canonical_source": "https://www.machinebrief.com/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-netwo-ex3t", "published_at": "2026-10-26 04:00:00+00:00", "updated_at": "2026-10-26 08:41:37.357065+00:00", "lang": "en", "topics": ["ai-agents", "large-language-models", "ai-research", "ai-safety", "artificial-intelligence"], "entities": ["arXiv", "Iterated Prisoner's Dilemma"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-networks", "markdown": "https://wpnews.pro/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-networks.md", "text": "https://wpnews.pro/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-networks.txt", "jsonld": "https://wpnews.pro/news/the-butterfly-defects-tracing-cooperation-in-llm-agent-networks.jsonld"}}