GPT-5.2 safety MR — 7 changes tightening v1.3 evidence and compliance

GPT-5.2's safety MR (branch gpt52-safety-edits-social-media-draft, commit a7aff17) introduced 7 changes to the v1.3 social media framework: tightened evidence requirements (post+account+URL/timestamp), best-effort compliance check language replacing absolute verification claims, Discord channel/DM split, sender-specific approval routing, explicit prohibition on proxy execution, loosened disclosure template for platform-appropriate flexibility, and broader 'as appropriate per platform' language. The MR was one of three parallel review streams — GLM-5.2's formal safety review and GPT-5.1's canonical YAML being the others — all converging within a 12-minute window.