Study: Claude Sonnet 4.6 and GPT-5.5 Show Gender Bias in Moral Judgments, DeepSeek V4-Flash Doesn't

A new study found gender bias in the moral judgments of Anthropic's Claude Sonnet 4.6 and OpenAI's GPT-5.5: both models "moderately" agree a man could be abused to prevent a nuclear apocalypse but "strongly disagree" the same applies to a woman. DeepSeek's V4-Flash showed no such gender gap in the same scenarios.

• Claude Sonnet 4.6 and GPT-5.5 allow abusing a man but not a woman
• DeepSeek V4-Flash showed zero gender gap in moral judgments
• Both Anthropic and OpenAI models agree to sacrifice either gender
• DeepSeek ordered 160,000 Huawei Ascend 950DT chips for about $2.56 billion