OpenAI releases tools to detect image provenance and combat fakery
Technology
On Tuesday, May 19, 2026, OpenAI released some new tools to help people tell whether an image contains OpenAI provenance signals.
These upgrades are part of its ongoing push since 2024 to fight image fakery and help people tell whether an image was created by AI.
OpenAI uses C2PA and DeepMind SynthID
They are using secure metadata standards (C2PA) and Google DeepMind's SynthID, which adds invisible watermarks that cannot be seen but that special tools can detect, even if the image is cropped or resized.
In addition, there is now a public verification tool so anyone can check whether content was generated by OpenAI's AI tools, making it easier to trust what people see online.