The AI just drafted a subject line for the Fieldnote relaunch and it is exactly as bad as you fear. The scan runs the banned list against the lowercased draft, in list order. What prints?
(Watch the order: the output follows the BANNED list, not the order the phrases appear in the draft.)