OpenAI is making it easier to check if an image was made by their models

OpenAI has introduced two new technical measures to improve the transparency of AI-generated content. By adopting the C2PA open standard for metadata and integrating Google's SynthID invisible watermarking, the company aims to provide more reliable provenance for images created by its models. While C2PA offers readable information about an image's origin, SynthID provides a robust, resilient watermark that persists through digital manipulation. OpenAI is also releasing a verification tool that allows users to check these signals, with plans to expand its capabilities to other AI tools in the future.

(Source：TechCrunch)