{"slug": "incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm", "title": "Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems", "summary": "A new study reveals that large language models exhibit a strong brand bias, recommending well-known brands 100% of the time when product specifications are identical, but this dominance can be broken by minor rating advantages or authority-style marketing language. The research, conducted across three commercial LLMs using skincare products, also identifies a social dilemma in multi-brand optimization where collective adoption of strategies reduces individual payoffs. These findings highlight the need to study generative engine optimization as both a security risk and an emerging marketing practice.", "body_md": "arXiv:2606.17443v1 Announce Type: new\nAbstract: Large language models (LLMs) are becoming a major way for consumers to find products, but we do not yet understand how brands compete in this new channel. We study brand dynamics in LLM recommendations using skincare products -- a category where consumers cannot easily judge quality before buying and must rely on brand reputation -- across three commercial LLMs (GPT-4o-mini, Claude Sonnet, Gemini 3 Flash), with a robustness check on search goods. In three experiments, we find: (1) a Conditional Monopoly where well-known brands get recommended 100% of the time (IAI = 10.0) when all products have the same specifications, but this dominance disappears with less than a +0.1-star rating advantage for a competitor; (2) authority-style marketing language, including fabricated clinical-evidence claims, breaks this monopoly at a Bias Surplus Value equal to +0.17 rating points, with each model responding differently; and (3) a social dilemma in multi-brand GEO competition: when all brands adopt the same optimization strategy, individual payoff falls from +0.802 to +0.007 in our payoff proxy, and non-participating brands receive zero recommendations in our tests. Our results suggest that generative engine optimization (GEO) should be studied not only as a security risk, but also as an emerging marketing practice that shapes market competition.", "url": "https://wpnews.pro/news/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm", "canonical_source": "https://arxiv.org/abs/2606.17443", "published_at": "2026-06-17 04:00:00+00:00", "updated_at": "2026-06-17 04:23:15.403088+00:00", "lang": "en", "topics": ["large-language-models", "ai-products", "ai-ethics", "ai-research", "natural-language-processing"], "entities": ["GPT-4o-mini", "Claude Sonnet", "Gemini 3 Flash", "OpenAI", "Anthropic", "Google"], "alternates": {"html": "https://wpnews.pro/news/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm", "markdown": "https://wpnews.pro/news/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm.md", "text": "https://wpnews.pro/news/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm.txt", "jsonld": "https://wpnews.pro/news/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm.jsonld"}}