OpenAI announces two steps aimed at helping people verify whether images are generated by its AI models. First, the company says it formally adopts the Coalition for Content Provenance and Authenticity (C2PA) open standard. C2PA is designed to provide provenance information that can be attached to digital content so downstream services and viewers can check a work’s origin and history.
Second, OpenAI says it is partnering with Google to embed SynthID, an invisible watermark, into its image outputs. SynthID is intended to be detectable through supporting tools, providing an additional signal that an image was produced by OpenAI’s systems.
Together, the measures are presented as complementary approaches—C2PA for provenance metadata and SynthID for an embedded identification mark. The announcements focus on improving transparency and making it easier to assess authenticity for AI-generated images, using open standards and Google’s watermarking technology.
