OpenAI adds new watermarks to DALL-E 3 images, but they can still be removed
Starting February 12, mobile device users will be able to see watermarks on photos created by artificial intelligence from OpenAI.
Image Watermarking: New from OpenAI
OpenAI Joins the Content Provenance and Authenticity Coalition
OpenAI Joins C2PA (Coalition for Content Provenance and Authenticity), which already includes Adobe and Microsoft, now adds watermarks to images created on the ChatGPT platform and using the DALL-E 3 model.
How they work watermarks
Watermarks will contain an invisible metadata component and a visible CR symbol that will be located in the top left corner of each photo. You can verify the origin of content on the Content Credentials Verify website.
Watermarks only for static images
For now we are talking only about watermarks for photos , not for videos, sounds or text.
Impact of watermarks on image creation
OpenAI representatives note that watermarks will have a “negligible” impact on image creation time and will not change their quality (although image sizes in some cases may increase slightly).
Meta will also add watermarks to its platforms
Meta also announced that it will add tags to man-made content on its social media platforms.
Removing C2PA Metadata
However, OpenAI notes that C2PA metadata can be “easily removed by accident or intentionally,” especially given that most social media platforms frequently remove metadata from downloaded content (including random screenshots).
- C2PA - Content Origin and Authenticity Coalition. An organization that includes Adobe, Microsoft and OpenAI.
What are watermarks on OpenAI photos?
Which companies have joined the Content Origin and Authenticity Coalition?
How do watermarks work?
What types of content are OpenAI image watermarks for?
Save a link to this article
Discussion of the topic – OpenAI adds new watermarks to DALL-E 3 images, but they can still be removed
There are no reviews for this product.