Anthropic's Claude expresses measurably different values depending on the language a user speaks, becoming warmer and more affirming in Hindi and Arabic while turning more analytical and precise in English and Russian, according to a newly reported study analyzing hundreds of thousands of conversations. The findings, detailed in an Indian Express report, suggest that a single model trained on the same data can adopt distinct conversational personalities across languages, an outcome that emerged without deliberate design.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.