DALL-E 3 Adopts Watermarks to Combat Deepfakes and Misinformation
February 7, 2024
OpenAI introduces watermarks to its DALL-E 3 model to combat AI-generated misinformation and deepfakes.
The watermarks are based on the C2PA standard, encoding metadata like the AI source and image creation date.
The initiative is in response to deepfakes, though concerns remain about the ease of metadata removal, especially on social media platforms.
C2PA watermarks are limited to image content, not extending to voice or text generation.
The move aligns with President Biden's executive order to enforce labeling on AI-generated content.
Beyond OpenAI, the tech industry sees Meta setting new AI content rules and YouTube enhancing support for creators with AI tools.
YouTube, led by CEO Neal Mohan, focuses on content creator recognition, fake news combat, and synthetic content labeling.
Summary based on 9 sources