My ChatGPT agent proofread a Google Doc I by no means requested it to open

However this explicit textual content from BenBit (to not be confused with BenBot, my home-grown agent) gave me a begin:

Small catch in your Publication 11-5 working doc: the teaser says Cantwell has a “three-point plan,” however the physique and her announcement describe six ideas. “Six-principle framework” would make them constant. I couldn’t confirm whether or not that teaser was utilized in a despatched model.

Uh, what?

BenBit was referring to a draft of my newest AI publication in my Google Drive, and as I shortly realized, not solely had BenBit seen the doc, it additionally opened it and proofread it, all with out my asking.

This caught me without warning, however earlier than I may dwell on what BenBit was doing in my Google Drive, I went into author mode: Had that discrepancy in my publication draft made it into the ultimate copy?

I quickly discovered why BenBit thought there was a mistake. I routinely depart dangling sections of rejected copy on the tail finish of my drafts, in order that contradictory “three-part” assertion was nonetheless sitting on the backside of my doc though I’d corrected the ultimate textual content. Confused by that enhancing quirk of mine, BenBit assumed I’d slipped up. No hurt, no foul. 

However whereas BenBit hasn’t really made any edits to my Google Doc, I used to be nonetheless startled that it had chosen to learn it over within the first place.