WE HAVE UPDATED OUR TERMS OF USE

We updated our TERMS OF USE. By continuing, you are agreeing to be bound by these updated terms.

  • What to Read Next
  • Most Popular News
  • Most Popular Opinion
  • Recommended Videos
You Have Been Selected
WSJ wants to hear from you. Take part in this short survey to help shape The Journal.

Advertisement

OpenAI Says It Can Now Detect Images Spawned by Its Software—Most of the Time

Startup’s new tool detects 98% of pictures generated by its DALL-E 3 system, but success drops if the images are altered

ET

0:00
Paused
0:04 / 10:38
Sora, OpenAI’s new text-to-video AI model, can create realistic scenes. The WSJ’s Joanna Stern sat down with the company’s CTO who explained how it works but ducked questions about how the model was trained. Photo illustration: Preston Jessee for The Wall Street Journal

AI is getting better at recognizing its own work. 

OpenAI on Tuesday is launching a new tool that can detect whether an image was created using the company’s text-to-image generator, DALL-E 3. OpenAI officials said that the tool is highly accurate in detecting DALL-E 3 images, but that small changes to a picture can confuse it—reflecting how artificial-intelligence companies are playing catch up in the ability to track their own technology.

Copyright ©2024 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8

Continue reading your article with a WSJ subscription.

Unlimited access across all platforms and devices.

WSJ Digital Bundle

$12.50/Week $7/Week

Includes unlimited access to The Wall Street Journal, Barron’s, and MarketWatch.

WSJ Digital

$9.75/Week $5/Week

Includes unlimited access to WSJ's unrivaled journalism.

Already a subscriber? Sign In

Continue reading your article with
a WSJ subscription

Subscribe Now

Already a subscriber? Sign In

Advertisement

Watch: China Launches Mission to ‘Hidden Side’ of the Moon
Watch: China Launches Mission to ‘Hidden Side’ of the Moon
02:37