OpenAI Adopts Google's SynthID Watermark for AI Images with Verification Tool
a day ago
- #Digital Trust
- #AI Transparency
- #Content Provenance
- OpenAI is strengthening content provenance with a multi-layered, ecosystem-driven approach to build trust online.
- They are conforming to C2PA standards to make provenance signals recognizable across platforms, ensuring metadata survives beyond initial creation points.
- A partnership with Google integrates SynthID invisible watermarking into images generated via ChatGPT, Codex, or the OpenAI API, complementing C2PA metadata.
- OpenAI is previewing a public verification tool to help users check if images contain provenance signals like Content Credentials and SynthID watermarks.
- The approach combines C2PA for detailed context and SynthID for durability against transformations, making provenance more resilient than either alone.
- No single method is foolproof; the verification tool cautiously avoids definitive conclusions if signals are stripped, and future updates aim to support cross-industry efforts.