{"ID":2862900,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.26464","arxiv_id":"2509.26464","title":"Extreme Self-Preference in Language Models","abstract":"Self-preference is a fundamental feature of biological organisms. Since large language models (LLMs) lack sentience, they might be expected to avoid such distortions. Yet, across 72 experiments and ~41,000 queries, we discovered massive self-preferences in eight widely used LLMs. In word-association tasks, models overwhelmingly paired positive attributes with their own names, companies, and CEOs over those of competitors. By manipulating LLM self-identification - revealing models' true identities or ascribing false ones - we found that preferences consistently followed assigned, not true, identities. Importantly, these effects were not explained by priming or role-playing and emerged in consequential settings, when evaluating job candidates and AI technologies. These results raise critical questions about whether LLM behavior will be systematically influenced by self-preferential tendencies, including a bias toward their own operation.","short_abstract":"Self-preference is a fundamental feature of biological organisms. Since large language models (LLMs) lack sentience, they might be expected to avoid such distortions. Yet, across 72 experiments and ~41,000 queries, we discovered massive self-preferences in eight widely used LLMs. In word-association tasks, models overw...","url_abs":"https://arxiv.org/abs/2509.26464","url_pdf":"https://arxiv.org/pdf/2509.26464v2","authors":"[\"Steven A. Lehr\",\"Mary Cipperman\",\"Mahzarin R. Banaji\"]","published":"2025-09-30T16:13:56Z","proceeding":"cs.AI","tasks":"[\"cs.AI\",\"cs.CL\",\"cs.LG\"]","methods":"[\"Large Language Model\",\"Language Model\",\"Generative Adversarial Network\"]","has_code":false}