How are errors made by an AI agent corrected?

A user signal, a scheduled review, or an automatic check can trigger a new verification. The team can correct data, change the publication status, or withdraw it. Relevant decisions and reviews are preserved to make the change auditable and to improve future controls.