Believe me, I get it–asking an AI chatbot to turn a picture of your pride and joy into a whimsical cartoon character is seriously fun. The appeal is undeniable, but yeah, there are risks.

This exact scenario actually played out at a family cookout a few weeks ago. When a well-meaning relative showed me an AI-altered family photo, my stomach dropped to my feet. I couldn’t help but think: Ah, crap… that photo’s just out there now, and who knows what could happen to it. I advised her to be more careful next time, but honestly? I’m not sure if my warning truly landed.

The problem here? It’s just plain unawareness. Pure, unfiltered unawareness.

Should you really be uploading that photo to a chatbot?

Please, just stop uploading photos of kids to chatbots. Or really, anyone who hasn’t said it’s okay. It might feel innocent, but there are real privacy risks here and not just for you. You might be giving up way more than you realize, and it’s easy to forget that when you’re just playing around.

Questions to ask before uploading any photo 

Before uploading any photos to your chatbot of choice, it’d be wise to have a conference with yourself and ask the following questions:

• Where’s this photo actually going?
• Could it be used to train the AI or shared without you knowing?
• Is there anything in it that gives away too much? (House number? Street sign?)
• Do you even know what the privacy policy says? (Be honest!)
• Did everyone in that photo say it was cool to upload?

What could go wrong 

I’m not trying to scare anyone here–I really do think AI can be helpful when used responsibly. But uploading personal photos can backfire if you’re not careful and paying close attention. Your photo shows way more than you think–think timestamps, location data, maybe even where you live. That kind of info is a straight-up goldmine to the wrong people.

Also, there’s the whole data breach risk. That means your photo might get leaked and used for sketchy stuff. If you’ve shared a selfie, for example, someone could easily turn it into a deepfake. In case you have no idea what the heck that is, it’s when AI superimposes your face onto someone else. It’s crazy easy to do and it can fool a lot of people and also have dire consequences.

Once you send off a photo, you’ve pretty much got no clue where it ends up or how it’s being kept–just because it disappears from the chat doesn’t mean it’s actually gone. There could be a copy of it sitting somewhere on a server. It might be used for training purposes, maybe for moderation, who really knows. That’s the scary part. You don’t really know who’s accessing your photos behind the scenes. 

You’ve got more control than you think

You’re not totally powerless here. A good place to start? Glance over the privacy policy and see what they’re actually doing with your stuff. A good (transparent!) one should provide clear answers to questions like…

• What kind of info are they grabbing? (Messages, photos, etc)
• How are they grabbing it?
• How long do they keep it?
• Where’s it being stored?
• Can you delete it?
• Can you opt out of being part of the training data pool?

OpenAI’s privacy policy covers many of the above questions. One thing you can do is turn off chat history in ChatGPT–that way, your conversations aren’t used to train the system. It’s a solid move, but yeah, not a 100 percent guarantee.

If you still want to share photos with a chatbot, you can actually strip the photo of its metadata. You can either use a third-party app like ExifTool or you can screenshot the photo in question, a process that automatically removes that information. 

Let’s talk about consent, because that one really matters to me.

Kids can’t give it. Period. I’m not sure why this is such a difficult concept for some people to grasp, but here we are. (Looking at you, family YouTube channels). Beyond the privacy issues, heavily altering your photos can have a seriously negative impact on how you might see yourself. Self-confidence can really take a nose dive here, especially if it’s an impressionable kid.

If you really want to mess around with AI, try using stock photos or AI-generated faces from This Person Does Not Exist. That way you’re not pulling from your personal library. 

Don’t take AI at face value

Chatbots sound human, but they’re not your friend (despite their often cheery disposition!). You can definitely have fun with AI, but just don’t take everything it says as gospel–it can make mistakes. You’ll want to keep an eye on your privacy and don’t be afraid to ask questions. Remember this: A bit of skepticism goes a long way in this day and age.

Author: Ashley Biancuzzo, Associate Editor, PCWorld