OpenAI announced major image watermarking update with C2PA and SynthID
Technology
On May 19, 2026, OpenAI announced a major update to its image watermarking technology, making it way harder for people to pass off AI-generated images as real.
By using C2PA provenance signals and SynthID-based digital watermarking, OpenAI's new system helps spot where an image really came from, even if someone tries to edit or mess with it.
OpenAI adopts SynthID and public verifier
OpenAI has become C2PA conformant and is incorporating Google DeepMind's SynthID technology, so its invisible watermarks stick around even after images are compressed or resized.
Plus, anyone can check if an image is legit using their new public tool at https://openai.com/research/verify/.
This upgrade is all about making digital content more trustworthy in a world full of AI fakes.