OpenAI Adopts C2PA Metadata and SynthID Watermarks for AI Images
OpenAI is joining the C2PA open standard and partnering with Google to embed invisible SynthID watermarks in its AI-generated images. The dual-layer approach pairs visible metadata with invisible watermarks, but it only covers OpenAI’s own products.
Announcement Highlights:
-
C2PA Standard Adoption: OpenAI is adopting the Coalition for Content Provenance and Authenticity (C2PA) open standard, which attaches metadata to files indicating their origin and any edits made. This standard has been ratified as an ISO and adopted by some Google products but remains inconsistent across the industry.
-
SynthID Watermarks: Google DeepMind’s SynthID technology embeds invisible watermarks directly into images, making them resistant to removal even through screenshots or other manipulations.
-
Complementary Systems: Both systems are designed to work together, with metadata providing contextual information and watermarking ensuring durability through image transformations.
-
Public Verification Tool: OpenAI is previewing a public verification tool that checks for both C2PA credentials and SynthID watermarks in images uploaded from OpenAI’s products. The company aims to expand its scope in the future.
Despite these measures, the scope remains limited to content produced by OpenAI’s tools, leaving much of the AI-generated imagery ecosystem without similar provenance standards.