Microsoft Paint and Photos Embed Invisible Watermarks in AI Images
The invisible watermark and remote moderation are not explained in Microsoft's stated C2PA manifest disclosure, raising a transparency gap in how AI-generated images are identified.
Reporting from 1 source: GIGAZINE.
Software developer and security researcher Xusheng Li found that AI image generation in Microsoft Paint and AI editing in Microsoft Photos send prompts to a remote server for moderation, which returns a GUID embedded as an invisible watermark. This happens even on Copilot+ PCs where generation works locally.
Microsoft's C2PA manifest explanation covers AI image identification, but it omits that the manifest carries a GUID received from a remote server, even when generation happens locally. Li's investigation showed prompt moderation is the trigger, and the watermark stays hidden unless the optional Copilot logo display is enabled.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.