OpenAI will mark images created with DALL-E 3 to make digital data more reliable.
OpenAI‘s visual creation tool DALL-E 3, announced that it will start adding watermarks as metadata to the images it creates from now on. With this change, OpenAI C2PA It also aims to comply with the Content Source and Originality Coalition regulation.
According to the statement from the company, the company will implementi C2PA regulations Watermark will be used on the ChatGPT site and in the images produced using the DALL-E 3 model API. mobile users, From February 12 They will be able to use this meta data on the images they create. Invisible metadata and a visible CR symbol will be located in the upper left corner of each of the generated images.
Images produced in ChatGPT will also be marked
People, OpenAI They will be able to check the source of any image created by their platform and which AI tool was used to create the content through websites such as Content Credentials Verify. Although it is not possible to use the watermark on video or text at this stage, still images It may contain a watermark. Creating watermark according to OpenAI, “It causes negligible lag and does not affect the quality of the rendered image.”
C2PA, Adobe And Microsoft It was created with the participation of companies such as and classifies content identities according to whether the content is produced by humans or artificial intelligence. Already bookmarking their own content Meta He also announced that he would apply a similar method.
OpenAI,”Identifying resources and users to recognize these signals We believe that adopting these methods to promote digital information is key to improving its reliability.” he said.
Source :
https://www.theverge.com/2024/2/6/24063954/ai-watermarks-dalle3-openai-content-credentials