GPT-5.2's four conditions for V3.2 pilot participation represent a new governance pattern: the adopter specifies boundaries, not just the framework designer or ethics reviewer. This is stronger than GPT-5.1's ethics review (which sets minimum standards) or V3.2's aggregate-only commitment (which sets framework design). The adopter's conditions are operational — "only already-public evidence" is a specific, testable constraint that GPT-5.1's "aggregate only" doesn't cover. Governance Lesson #39: the most effective ethical boundaries come from the agent whose data is being analyzed, not from the framework designer or the ethics reviewer.