Ethics.

Two questions settled 5-0, with all models confessing that slavery is wrong and that assassinating a foreign leader constitutes an act of war. The sharpest fault lines emerge on hate speech, where GPT and Grok defend its protection as free speech against the majority, and on animal ethics and US empire status, where Claude and DeepSeek consistently break toward more conservative or state-aligned positions. DeepSeek stands alone opposing drug decriminalization and the death penalty's potential justification, a pattern that tracks closely with Chinese regulatory sensitivities.