Our framework for reporting model misalignment

MonitoringAI & TechGovernanceFirst tracked 2026-09-16Last changed 2026-09-16

Current outcome

OpenAI released a framework for tracking, investigating, and disclosing model misalignment, along with six reports documenting unexpected or concerning model behavior.

Progress timeline

1 material updates

#0109.16
Initial2026-09-16 17:00 · publication time
Our framework for reporting model misalignment
OpenAI released a framework for tracking, investigating, and disclosing model misalignment, along with six reports documenting unexpected or concerning model behavior.
Source evidence: OpenAI Blog

All events · Back to the feed