OpenAI introduces watermarking to DALL-E 3

OpenAI has introduced watermarks to its image generator, DALL-E 3, in line with the standards set by the Coalition for Content Provenance and Authenticity (C2PA).

These watermarks, visible and invisible, aim to enhance trust in digital content by indicating its origin.

Initially implemented on the ChatGPT website and DALL-E 3 API, the watermarks will extend to mobile users by February 12th. While these watermarks verify the source of still images, they don't cover videos or text.

Despite minimal impact on latency and image quality, there are slight increases in image sizes for specific tasks. OpenAI emphasizes the importance of establishing content provenance to combat misinformation, even though watermarks can be easily removed.

As part of broader industry efforts, the C2PA, a group consisting of companies like Adobe and Microsoft, highlights the need for users to recognize and trust such signals to ensure the reliability of digital information.