OpenAI is expanding ChatGPT’s e-commerce tools by introducing a virtual try-on feature that allows users to visualize how clothing items might look on them. The function operates similarly to a tool Google launched in 2025, enabling shoppers to upload personal photos and see themselves wearing various garments through AI-generated imagery.
Users can access the feature directly within product listings displayed in their chat threads. By selecting a clothing item and tapping “Try on,” individuals can upload both a selfie and a full-body photo for the system to process. Additionally, users are not limited to products surfaced by the chatbot; they can upload screenshots of any clothing item to generate a virtual fitting image.
The company states that the new imagery is produced using the ChatGPT Images 2.5 model, which was released in September. Alongside the try-on capability, OpenAI has added a library feature within the app, allowing users to save products for future reference. However, the company reiterated its standard data privacy warnings: images uploaded from personal accounts may be used for training purposes unless users explicitly opt out.
Although OpenAI does not operate a marketplace comparable to Amazon, it has begun generating revenue from shopping-related activities through its Instant Checkout feature, which takes a cut of certain transactions. The platform also hosts ads for merchants. While these income streams may not currently offset the billions OpenAI spends on AI infrastructure, the steady introduction of shopping features over the past year suggests the company views this sector as a potential long-term revenue driver.
Google already did this in 2025. Kind of surprising OpenAI is just catching up now with this particular feature.
Are we sure about the data privacy warnings? Using my personal photos for model training is a huge red flag for me.
Finally, a way to actually see if those trendy pants fit. This feels like a game changer for online shopping frustration.