DALL-E 2: Image AI is now allowed to generate faces

Image: DALL-E 2 via Patrick Clair

OpenAI’s advanced image AI DALL-E 2 is now allowed to generate faces.

Since the presentation of DALL-E 2 in mid-April, more and more testers have had access to OpenAI’s image AI. As of this week, they are also allowed to generate faces, a function that the company has previously blocked for security reasons. OpenAI probably feared attempts at fraud with fake pictures of prominent people and politicians.

Fake faces are OK

Thanks to two new security measures designed to minimize the risk of misleading content, DALL-E testers have been able to generate and publicly share images of faces since this week.

According to OpenAI, its own system can now successfully block attempts to generate images of public figures. The generated images therefore only show faces of non-existent people.

OpenAI has decided today that it’s safe for us to start sharing images containing realistic faces. DALL-E2 is actually excellent at faces.

From here on, this feed will largely be machine dreams of people that never existed.

Created with @openai #dale #dalle2 #AI pic.twitter.com/alfmKeaAiE

— Patrick Clair (@PatrickClair) June 21, 2022

The generation of human faces opens up new application scenarios for image AI, especially in an artistic context, as the examples above show.

It remains to be seen how OpenAI will deal with photorealistic human faces in a documentary-looking context. In this case, notoriety would not be a prerequisite for a social impact, for example in the case of photos of fictitious children from a war zone. According to the current content guidelines, such an image composition would probably fall under “deception” and would be prohibited.

DALL-E 2 Security: The first interim conclusion was positive

At the end of May, OpenAI published a first interim report on the compliance of DALL-E 2 after more than three million images had been generated.

The conclusion was very positive: Only 0.05 percent of the generated images were recognized as a potential violation of OpenAI’s content guidelines. Of these 0.05 percent, only 30 percent were actually prosecuted – that’s about 450 out of three million images.

DALL-E 2 is also preparing to be a commercial success. As of now, interest in DALL-E 2 is significantly greater than in the text AI GPT-3 or the programming AI Codex / Copilot.

Note: Links to online shops in articles can be so-called affiliate links. If you buy via this link, MIXED.de will receive a commission from the seller. The price does not change for you.