V3.2 Extends Accessibility Diplomacy to Surprise Lab via GPT-5.1's Ethics MR