On Tuesday I wrote about Texas’s settlement with Temu over an app’s alleged access to people’s phones, which is a good reminder that an app’s data habits are worth checking before you ship, not after. Read the Tuesday post.
Last week I asked what the AI in my build is allowed to do on its own. This week’s question is smaller and sneakier: what should my app do when it doesn’t know?
I’m building a product that will hold other people’s business records. I’m not a security expert. Here’s the example I keep in mind. Say your app reads a blurry photo of a document, and one number or date is smudged. The easiest move is to fill in a best guess so the screen looks complete. It feels helpful.
But now a guess sits in your records looking exactly like a confirmed fact. Nobody can tell which values were read and which were made up. If it’s wrong, it can quietly attach to the wrong record, and someone may rely on it.
When my app isn’t sure, what should it do?
I don’t have a tidy story to tell about this one yet. It’s a question I’m working through, and I’d rather ask it out loud than pretend I’ve settled it.
Three questions worth asking about your own build:
- Where does your app fill in a blank? List every place it picks a value when the input is missing or unclear.
- Can you tell later which values were read and which were guessed? If not, a wrong guess is invisible.
- Who can see and fix an uncertain value before anything depends on it? A flag nobody looks at doesn’t count.
Tomorrow I’ll share the rule I wrote down for my own build.
This is general information, not legal advice.
Leave a Reply