DALL-E 3 is Getting New Watermarks from OpenAI

**OpenAI’s DALL-E 3 Adds Watermarks to Image Metadata**

OpenAI’s image generator DALL-E 3 will now add watermarks to image metadata as it integrates support for standards from the Coalition for Content Provenance and Authenticity (C2PA). This move comes as more companies roll out their support for the C2PA standards.

The company announced that watermarks from C2PA will be visible in images generated on the ChatGPT website and the DALL-E 3 model’s API. Mobile users can expect to see the watermarks by February 12th, and they will include both an invisible metadata component and a visible CR symbol in the top left corner of each generated image.

To check the provenance of any image generated by OpenAI’s platforms, people can use websites like Content Credentials Verify. However, at present, only still images, and not videos or text, are eligible for the watermark.

OpenAI reassures users that adding the watermark metadata to images will have a “negligible effect on latency and will not affect the quality of the image generation.” It may, however, slightly increase image sizes for some tasks.

The C2PA, a group made up of companies such as Adobe and Microsoft, is behind the push for the use of the Content Credentials watermark to identify the provenance of content, as well as to indicate if it was created by humans or with AI. Adobe has created a Content Credentials symbol, which OpenAI is now including in DALL-E 3 creations. Meta has also announced that it will add tags to AI-generated content on its social media platforms.

Identifying AI-generated content is a core directive in the Biden administration’s executive order on AI. However, OpenAI points out that C2PA’s metadata can “easily be removed either accidentally or intentionally,” especially given that most social media platforms often remove metadata from uploaded content. Taking a screenshot also omits the metadata.

“We believe that adopting these methods for establishing provenance and encouraging users to recognize these signals are key to increasing the trustworthiness of digital information,” OpenAI says on its site.

In summary, OpenAI’s DALL-E 3 model will now include watermarks in its generated images to establish the provenance of content. The move aligns with a broader industry trend towards utilizing watermarks to identify the origin of content. However, concerns about the effectiveness of such watermarks in stopping misinformation persist, as they can be easily removed or omitted from uploaded content.

For more information, visit GPTNewsRoom.com.

*Please note the link provided has a nofollow tag.

from GPT News Room https://ift.tt/GwWum4L