OpenAI upgrades invisible watermarking to verify provenance of AI images
Technology
OpenAI just made it easier to spot AI-generated images. It has upgraded its watermarking technology so you can check where an image really came from.
The new system uses invisible watermarks that stay put even if the image gets resized or compressed, fixing earlier issues where people could easily remove metadata.
OpenAI launches public SynthID verification tool
This upgrade uses Google DeepMind's SynthID to embed those hidden marks, and there is now a public tool on OpenAI's website so anyone can verify if an image was made by its tools.
OpenAI says this is all about being more transparent and keeping digital content trustworthy, thanks in part to C2PA conformance.