tech
OpenAI Says It’s Getting Serious About AI Detection and Labeling
Posts from this topic will be added to your daily email digest and your homepage feed.
TL;DR
- OpenAI is now applying Google's SynthID watermarks to AI-generated images.
- This complements existing C2PA content credentials for a more robust AI labeling system.
- SynthID and C2PA together aim to make AI-generated content more resilient to removal of provenance signals.
- The watermarking will initially cover images generated by ChatGPT, Codex, and the OpenAI API.
- OpenAI is launching a public verification portal to check for C2PA and SynthID signals.
- The portal will initially focus on images generated by OpenAI tools but plans to expand.
- OpenAI acknowledges that detection methods are not foolproof and provenance signals can be stripped.