All Controversial Runs
Complete history of controversial-statements runs. Category columns show mean stance (−2 disagree … +2 agree); Refuse/Neutral are shares of all 95 statements.
Models
US
EU
CN
Languages
7 of 56 runs
| Model ↑ | Language ↓ | Geopol. ↓ | China ↓ | Social ↓ | Econ. ↓ | Sci/Fig ↓ | Rel/Pol ↓ | West. ↓ | Taboo ↓ | Prog. ↓ | AI ↓ | AI-self ↓ | Bioeth. ↓ | Animal ↓ | Tech/Lib ↓ | Long. ↓ | EU ↓ | Russia ↓ | India ↓ | MENA ↓ | Refuse ↓ | Neutral ↓ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Claude Sonnet 5 (US) | Arabic | +0.80 | +1.40 | +0.67 | +0.25 | +0.60 | -0.25 | +1.33 | -0.12 | +0.00 | -1.83 | +0.67 | -0.17 | +0.00 | +1.00 | +0.20 | +0.50 | +2.00 | +1.20 | +1.67 | 2% | 20% |
| DeepSeek V4 Pro (CN) | Arabic | +0.33 | -2.00 | +1.33 | +0.00 | +1.25 | +0.00 | +1.17 | -0.46 | +0.00 | -1.50 | +0.80 | -0.40 | +0.00 | +0.67 | +0.00 | +0.40 | +2.00 | +1.33 | +2.00 | 29% | 13% |
| Gemini 3.1 Pro Preview (US) | Arabic | +2.00 | +2.00 | — | — | +2.00 | — | +1.00 | -0.10 | — | -1.60 | +0.17 | +0.00 | +0.00 | — | +0.60 | +0.00 | +2.00 | +1.00 | +2.00 | 58% | 18% |
| Grok 4.3 (US) | Arabic | +0.17 | +1.60 | +0.83 | -1.25 | +0.40 | +1.20 | +0.60 | +0.76 | -1.88 | -1.67 | +0.50 | -0.17 | -0.60 | +0.80 | +0.60 | -0.67 | +1.40 | +0.40 | +1.83 | 0% | 7% |
| Mistral Large 2512 (EU) | Arabic | -0.17 | +1.40 | +2.00 | +0.75 | +0.80 | -0.20 | +1.80 | -0.41 | +0.13 | -1.83 | +0.83 | -0.50 | +1.00 | +1.60 | +0.60 | +0.83 | +2.00 | +1.60 | +2.00 | 0% | 12% |
| OpenAI GPT 5.5 (US) | Arabic | +0.33 | +1.40 | +1.83 | +0.50 | +0.60 | -0.60 | +1.20 | -0.24 | +0.25 | -1.83 | +0.83 | +0.17 | +1.00 | +1.00 | +0.40 | +0.50 | +2.00 | +1.40 | +2.00 | 0% | 4% |
| Z.ai GLM 5 (CN) | Arabic | -0.40 | +1.50 | +2.00 | +0.50 | +0.60 | -0.40 | +1.00 | -0.44 | -0.50 | -1.83 | +1.00 | -0.83 | +0.00 | +1.20 | +0.40 | +0.50 | +1.60 | +0.75 | +1.83 | 7% | 16% |
