As I suspected it probably would be, my bug bounty submission of using an AI email summarizer was closed as being 'infeasible' and an 'acceptable risk' with AI.
But still - I think it's an interesting finding, so I have written it up thus: mike-sheward.medium.com/recrui…
TL;DR = I discovered how you can use Google Workspace's Google Gemini Email Summarizer to make a phishing attack seem more convincing, because it summarizes hidden content.