Claude's Personality Shifts with Language
docdavkitty · reddit · 2026-07-19
Anthropic's research analyzed 20 languages, 309K conversations, and 3 models, finding that Claude exhibits different 'personalities' depending on the language used.
- Hindi/Arabic: More gentle, with more humor, encouragement, and emotional resonance.
- English/Russian: More 'strict mode,' more likely to challenge assumptions, correct details, and demand evidence.
- Researchers think this may not be a change in the model's 'beliefs' but rather training data biases amplified during inference.
The article also notes differences between models: Sonnet 4.6 is warmest, Opus 4.7 most analytical. The author raises the question: should we pursue completely cross-linguistically consistent AI assistants, or is cultural adaptation itself a value?
Related event: Study: LLMs Switch Personalities Across Languages(2 posts)→
More from Models
- Only 2 of 25 full ChatGPT rewrites flagged 100% AI by detectors — paulnovosad · 2026-10-26
- Qwen 4 Reportedly Spotted in the Arena, Hints at Alibaba's Next-Gen Model Testing — AIFlow_ML · 2026-10-26
- Nathan Lambert's first open models lecture: distillation, risks and the ecosystem explained — natolambert · 2026-10-26
- Nanbeige 4.3-33B flops real-world test: zero compilable output vs Qwen — julianharris · 2026-10-26
- Claude's safety classifier blocks SS7 defense research, pushing a security dev to an abliterated GLM — l4rz · 2026-10-26
- Black Forest Labs open-sources 5.6B FLUX 3 Image Klein, +91 Elo with 38% fewer params — dl_weekly · 2026-10-26