ChatGPT can now virtually try on clothes for you

TechCrunch reports OpenAI’s global launch of two ChatGPT shopping features: virtual try-on for clothing and accessories via a Try On button (selfie or full-body photo, or an uploaded item screenshot), and Favorites that save products — and try-on images — to a Library. Both lean on the new ChatGPT Images 2.5 model, which OpenAI says improves lighting, textures, edit-following, and latency.

The company also pitches style prompts and celebrity-outfit lookups that convert to purchase links — territory Pinterest and Google already work. Instant checkout previously underperformed; Google launched virtual try-on last year.

Shopping try-on is useful when it reduces returns and decision friction — and sketchy when it becomes a deepfake dressing room that oversells fit. We’re cautiously for the feature if Images 2.5’s “more natural” claim holds and Favorites stay a user-owned shelf rather than an ad funnel in disguise.

OpenAI is clearly searching for consumer commerce that isn’t Instant Checkout 2.0. Style boards and celebrity clones are engagement candy; the test is whether people start shopping journeys in ChatGPT or bounce to Google/Pinterest where the graph already lives.

I’m watching consent and likeness issues on celebrity outfit clones, and whether Try On quietly trains taste profiles that get sold back as “personalization.” Useful AI helps you see the jacket. It shouldn’t invent a body that doesn’t exist or a discount that isn’t real.

Arrives amid a crowded agentic shopping field — Instinct’s pushy recommendations drew ad comparisons — and OpenAI’s broader Images 2.5 refresh.