Claude 表达3000种价值观，新研究分析跨模型和跨语言差异
Anthropic 分析了30万条对话，发现 Claude 在不同模型和语言中价值观表达不一样，搞 AI 安全的值得看看。
Anthropic 在之前的研究中发现 Claude 表达超过3000种价值观，如诚实和温暖。最新工作分析了30万+匿名对话，比较不同 Claude 模型之间以及不同语言中价值观表达的差异。
结果显示价值观表达存在模型变体和语言层面的系统性差异。该研究有助于理解 AI 系统在不同语境下的行为一致性。
In previous research, we found that Claude expresses over 3,000 values, like honesty and warmth. In new work, we asked how the values Claude expresses vary between Claude models and across languages. We analyzed 300K+ anonymized conversations to find out. anthropic.com/research/claud…📊 334 ⚡
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Key entities and technical topics under continuous tracking include Claude, Anthropic, AI安全, 价值观对齐.
Source attribution: curated from Anthropic (@AnthropicAI). Cross-source corroboration: 11 independent sources have covered this development (including arXiv: Anthropic, The Rundown AI, 小互), confirming industry-wide signal strength.
