Dall-E 2 users to be allowed to upload faces for first time

Dall-E 2 users to be allowed to upload faces for first time

Feature marks latest relaxation of rules around how image-generating AI tool can be used

Dall-E 2 on a phone screen

Users of the image generating artificial intelligence Dall-E 2 will be allowed to upload faces to the system for the first time, creators OpenAI have said, as competition in the sector heats up.

The feature marks the latest relaxation of the company’s rules around how its tool, which can generate high-quality images from a text prompt, can be used. When it first launched in a public beta, OpenAI banned users from generating any images with a realistic face.

Later, those rules were relaxed to allow the generation of realistic faces, but not those of specific individuals.

Now, users will be able to upload photos that depict real people – with consent – and use OpenAI’s tools to generate new variations on the pictures. Twitch streamer NymN, for instance, used the technology to generate images of himself as a WWE wrestler taking a bath.

Nice man Dall-e is allowing faces again. He is me as a wwe wrestler taking a bath pic.twitter.com/bwoCHIDylF

— NymN (@nymnion) September 19, 2022

Manuel Sainsily, an artist who works with AI tech, shared a more prosaic use for the feature: generating endless hats.

🤖 AI Image Variations x #OutPainting 🤯

🧢 None of these hats or sunglasses exist in the original photo. No cap.

✨ Editing photos with faces is back on @OpenAI, and the possibilities are endless!

🙏 Inspired by the talented @MichaelCarychao pic.twitter.com/GSZ0HtsQLg

— Manu.Vision | Futurist (@ManuVision) September 20, 2022

In a letter to users, OpenAI explained its rationale for softening the rules. The company had improved the technology it used to prevent users from generating sexual and violent content, it said, and decided that the trade-offs were now worth it: “Many of you have told us that you miss using Dall-E to dream up outfits and hairstyles on yourselves and edit the backgrounds of family photos.

“A reconstructive surgeon told us that he’d been using Dall-E to help his patients visualise results. And filmmakers have told us that they want to be able to edit images of scenes with people to help speed up their creative processes.”

But OpenAI is also reacting to pressure from the competition, particularly London-based Stability.AI, whose Stable Diffusion image generation model was released to the public with effectively no moderation or restrictions at all.

“Ultimately, it’s people’s responsibility as to whether they are ethical, moral, and legal in how they operate this technology,” Stability.AI’s chief executive and founder, Emad Mostaque, told the Verge earlier this month.