OpenAI is adopting Google DeepMind’s SynthID technology to identify AI-generated images. The company is also conforming to the C2PA open standard for content provenance. This dual strategy increases the resilience of content origin information.
SynthID embeds invisible watermarks directly into image pixels. The technology remains detectable despite modifications such as screenshots, compression, or cropping.
The C2PA standard provides a verifiable history of content creation and modification. OpenAI is also launching a public tool to verify if its models generated specific images.