One of the things I hate about AI is how it makes everyone lower their standards with regard to user privacy.
Including companies like Apple that have a well deserved reputation for protecting it.
Yesterday I got a push notification about a new feature in the Feedback app that everyone in the iOS beta has installed. There’s a new “Rate Your Experiences”.
Which is a cleverly named function that sends your and your friends private data to Apple in plain text.
There’s a new purple button:
When you tap on that button, you get a short overview:
You are reassured that everything stays on device, but as we’ll shortly see, that’s a lie.
Now I’m curious, how do you improve AI responses without the original plain text that generated them?
The next screen tells me I can rate with thumbs up or down. Seems reasonable.
And earn points - this is when my horseshit detectors went to 11.
You are then presented with personal messages with the Apple Intelligence.
For example:
When you do a thumbs down, Apple wants the data.
This is where it goes off the rails and becomes a privacy clusterfuck.
This is the next screen:
9 messages from friends are attached to the response AUTOMATICALLY.
It took me awhile to even figure out how to remove them (they, like other FB reports have attachments, and they can be deleted).
Most people are just going to press the big purple button.
Craig Hockenberry
Owen 🇦🇺