OpenAI is working on new AI image detection tools. The recognized startup artificial intelligence has incorporated a new tool to identify if an image was created with its DALL-E image generator, in addition to introducing advanced watermarking methods to more obviously mark the content generated by its systems.
In a recent article, OpenAI announced that it has started developing new methods to crawl content and check if it was generated by AI. This includes a new image detection classifier — which actually uses AI — to determine if the photo was generated by AI, as well as a tamper-resistant watermark that can tag content such as audio with invisible cues. The company expects these watermarks to be difficult to remove.
OpenAI said its new detector could correctly identify 98.8% of images created by DALL-E 3
OpenAI said its new detector could correctly identify 98.8% of images created by DALL-E 3, the latest version of its imager. But the company said the tool was not designed to detect images produced by other popular generators such as Midjourney and Stability.
Like tech giants Google and Meta, the company will join the steering committee of the Coalition for Content Provenance and Authenticity, or C2PA, an effort to develop credentials for digital content. The C2PA standard is a kind of “nutrition label” for images, videos, audio clips and other files that shows when and how they were produced or altered, including with AI.
According to Open AI, people can still create misleading content without this information (or can remove it), but they cannot easily falsify or alter it, making it an important resource for building trust.
“To drive adoption and understanding of provenance standards, including C2PA, we teamed up with Microsoft to launch a social resilience fund. This $2 million fund will support education and understanding of AI, including through organizations like AARP’s Older Adult Technology Services. , International IDEA and Association on AI,” the company said.
