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.https://t.co/PgxsMXipt5
Anthropic's cross-model, cross-language analysis of Claude's expressed values helps engineers anticipate behavioral drift when switching models or working in non-English contexts — useful signal for model selection and safety-sensitive deployments.
