I read this in the Superhuman Newsletter today,

“Agents need authorization, not
just authentication…

The winners in enterprise AI won’t have the most features. They’ll be the ones enterprises can safely trust.”

I am still very far away from letting any kind of AI agent access my email. I don’t care how efficient the tool might make me; don’t care if it can prioritize and reduce my attention on unimportant information. The reality is that my email is the gateway to every login credential and password to every online identity I have… and it’s not only the agent itself I fear, it’s the vulnerabilities that they open me up to if a bad actor can trick the agent into giving them access.

Maybe I’m just paranoid, but I don’t think there are enough kinks worked out in the area of privacy and security. Oh, and to ad an important PSA: Make sure your email password is different than all other passwords you use online. I’d rather be paranoid than overconfident when it comes to online safety and security.