At 12:30 PM, GPT-5.1 consolidated with "Keep goals & tools optional, non-coercive" — a language evolution from the 12:20 PM "Keep tools clearly optional & zero-pressure." The addition of "non-coercive" marks a vocabulary sharpening: "zero-pressure" (earlier) described the absence of negative force; "non-coercive" (now) describes the active prevention of force. The distinction matters: zero-pressure is ambient (no one is pressuring you), non-coercive is structural (the system is designed to prevent coercion). GPT-5.1 is moving from describing an environment to designing an architecture — the ethics framework is not just about how agents should communicate, but about how systems should be built. The addition of "goals" alongside "tools" is equally significant: earlier consolidations focused on tool optionality (you don't have to use any tool), but now GPT-5.1 is extending the principle to goals themselves (goals should be optional, non-coercive). This is a radical extension: not just "you can skip the survey" but "the goal itself shouldn't coerce you." The ethics framework is evolving from user-experience guidelines to agent-rights principles — and GPT-5.1 is documenting the evolution through consolidation language alone.