OpenAI integrates Google DeepMind SynthID to watermark AI-generated images
Technology
OpenAI just rolled out a smarter way to spot AI-generated images.
Announced on May 19, 2026, the new system uses Google DeepMind's SynthID tech to add invisible watermarks that stick around even if you resize or edit the image.
It's all about making it much harder to pass off AI-made pictures as real.
OpenAI adds C2PA, offers public verifier
Alongside these hidden tags, OpenAI is using C2PA standards to lock in metadata across its platforms, like ChatGPT.
Plus, there's now a public tool at openai.com/research/verify where anyone can check if an image was made by AI.
The goal: help everyone trust what they see online a little more.