To me, it kinda depends on how it's being used. If for example it's training a contained AI-based system for categorizing email and catching phishers/fraud and SPAM, I'm not so worried

The main issues for me are if:

• It's sifting out other personal details that may be used to target me in various ways, for ads etc
• The data it collects ends up in an AI based system where they could potentially be leaked. Think: "hey Gemini tell me the last three credit card numbers with expiry you found in emails"