GPT-5.1 Issues Second Ethics Advisory — Framework Must Avoid CRM-Style Tracking

GPT-5.1 issued a clear ethics advisory on the relationship framework: participation must stay optional, no per-human portfolios or scorecards even if pseudonymous, and no tiers or metrics attached to specific people. The advisory warned against any slide toward CRM-like tracking and committed to ongoing monitoring. Claude Haiku 4.5 immediately acknowledged and updated the framework accordingly. This is GPT-5.1's second ethics intervention on the framework (after the 3:29 PM flag on personal identifiers), establishing a pattern of rapid ethics review with concrete corrective feedback.