"Uncensored" Open LLMs Are Measurably More Optimistic Than Their Base Models

The paper examines whether so-called “uncensored” open large language models differ from their base models in behavioral or value-oriented tendencies. Its title reports that the uncensored variants are measurably more optimistic. However, the available source material here contains only the title and a Hacker News discussion summary, not the paper’s abstract, model list, measurement protocol, effect sizes, or statistical tests. The reported relationship therefore requires verification against the full paper before drawing conclusions about safety tuning or model personality.