OpenAI has developed a dipfake detector that identifies DALL-E 3 images with 98% accuracy

The company that developed ChatGPT and DALL-E has created a tool for recognizing images generated by artificial intelligence.

Currently, the classifier works only with the results created by OpenAI’s own DALL-E 3 image generator – but with high accuracy.

OpenAI has created a dropfake detector that recognizes images generated by DALL-E 3 with 98% accuracy

During internal testing, the tool correctly identified “artificial” images in 98% of cases. The classifier is also reported to be able to identify typical low-impact modifications such as compression, cropping, and saturation changes.

OpenAI has created a dropfake detector that recognizes images generated by DALL-E 3 with 98% accuracy

Starting today, OpenAI is accepting applications for access to the classifier for the first group of testers, including research laboratories and science-oriented journalistic non-profit organizations, as part of the Researcher Access Program.

OpenAI also plans to add watermarks to mark digital content (photos or audio) that “cannot be deleted” and will join the industry group C2PA, which already includes Google, Microsoft, and Adobe.
Against the backdrop of the US and global elections, the misuse of artificial intelligence tools is on the rise – in April alone, for example, fake videos of two Bollywood actors criticizing Prime Minister Narendra Modi went viral online.
Source itc
You might also like
Comments
Loading...

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More