OpenAI Adds New Watermark to DALL-E3 to Enhance Digital Information Credibility
-
As more companies support the Content Authenticity Initiative (C2PA) standard, OpenAI's generator DALL-E3 will add watermarks to images as metadata. These watermarks will appear in images generated by the ChatGPT website and the DALL-E3 model's API, helping users verify the source and creation tools of the images.
Mobile users will see these watermarks by February 12, which will include invisible metadata components and a visible CR symbol. Although watermarks are not absolutely secure, they can provide a certain level of support for the credibility of digital information. C2PA is an organization formed by companies like Adobe and Microsoft, dedicated to promoting the use of content watermarks to identify the source of content and indicate whether it was created by humans or AI. OpenAI is actively involved in this initiative, hoping to enhance user trust in digital content through the addition of watermarks.
By embedding watermark metadata into images, OpenAI states that the impact on image generation quality is minimal, and for certain tasks, it might slightly increase the image size. While watermarks can help determine content origins and encourage users to recognize these signals, OpenAI also notes that C2PA's metadata can be easily removed, either intentionally or unintentionally, especially on social media platforms. These clues are a step forward, but ensuring the credibility of digital information still requires more tools and methods.
Ultimately, these watermarks can help users better understand AI-generated content, improving the trustworthiness and transparency of digital information. As technology continues to evolve, we can expect to see more protective measures and innovative solutions regarding the source of digital information.