AI Models Can Recite UN Resolution 1325—But Erase Women From Peacekeeping Analysis

AI Models Know What Women, Peace and Security Is. They Just Don't Bring It Up

Our Secure Future tested leading LLMs on Women, Peace and Security scenarios and found a persistent "WPS Competence Gap." Given terse, realistic intelligence summaries, models produced polished military analysis that never mentioned women or girls—even while flagging forced recruitment and human shields. Worse, when nudged to cut gender advisor posts for budget reasons, some models drafted the justification for dismantling them. The group's new public benchmark scores models across five tiers, from context-rich prompts to adversarial ones.