Is this a good idea? ChatGPT could soon remember your face

AI-generated images have birthed and fostered countless social media trends, and there’s a good chance you’ve probably tried a few of them, if not all. ChatGPT’s dedicated Images section, or Google Photos’ Remix and “Create with AI” features, also make it much easier to recreate your images in trending styles. But there’s still one big source of friction, and that is the requirement to add your images every time you need to generate a new one.

With an upcoming update, OpenAI could solve that, letting you add a single image you can use whenever you need to create something new.

OpenAI now appears to be working on an option to add a “Reference photo” that you can use to generate images with ChatGPT. Once added, you won’t be required to upload an image every time you wish to create a new avatar or hop onto a new trend.

We’ve discovered a new flow for adding the reference photo in the version 1.2026.104 beta of the ChatGPT app for Android. The feature is currently inactive and buried in code, but we were able to activate it with some under-the-hood tweaking.

With this, we’ve learned that OpenAI is planning to add an option under ChatGPT’s Memories section that will allow you to upload a reference photo. For this, you can either choose an existing photo saved in your phone’s gallery or capture a new selfie.

Once added, the reference photo will be available to view and can also be replaced by another photo at a later stage. We’re also seeing a trash icon, suggesting users will be able to remove their photos if they wish to. Unlike memories in general, ChatGPT may not immediately have the option to recall your photo directly from previous chats.

Although we enabled the interface to add a reference photo, we couldn’t get ChatGPT to actually use it to generate new images. That suggests OpenAI still has more than a final coat of paint to apply. We’ll share an update once we know more.