The Internet Was Ghibli-fying Everything—Until OpenAI Ruined the Fun

OpenAI couldn’t even let us enjoy their technology for more than a day.
Yesterday OpenAI dropped its new “Images in ChatGPT” feature, an upgrade to GPT-4o that allowed users to upload images directly in the chat and alter them to impressive degrees, and for a brief, glorious moment, it felt like finally, we can have fun with AI again.

This wasn’t DALL·E 3 stapled on the side; with boring, steral, uncanny images, this was something new. GPT-4o, OpenAI’s shiny “omnimodal” model, was suddenly doing text, images, audio, and video in one sleek, conversational interface. It felt like the most powerful version of the AI future we’ve been promised.

The internet as if in one unified mind began Ghibli-styling everything. And then, true to form, OpenAI did what they do best: panic, handcuff their own tech, and nerf the magic before it could even finish trending.

Because if there’s one thing OpenAI has mastered, besides building cutting-edge tools, it’s being absolutely afraid of their own technology. Every time they develop something new, a suit in a corner office gets sweaty, and the whole thing gets yanked, throttled, or shoved behind a velvet rope.

Here are the best Studio Ghibli creations we could find before OpenAI turned the fun off.