Assistive AI should have a configurable rule that automatically ignores email banners, headers, disclaimers, and security warnings when determining priority or categorization. These banners often contain language that triggers higher urgency levels or incorrect categories, even though they are not part of the actual user request. The AI should base its prioritization solely on the substantive content of the email, unless the sender explicitly reports a real security incident. This rule should be applied universally across all prioritization and categorization logic to prevent misclassification caused by banner text.