Watermarking and Metadata for GenAI Transparency at Scale - Lessons Learned and Challenges Ahead

The proliferation of generative AI (GenAI) technology has revolutionized content creation across various online platforms. This advancement has sparked significant public debate concerning the transparency around AI-generated content. As the difference between human-generated and synthetic content gets blurred, people want to know where the boundary lies. Invisible and visible watermarks, content labels, and IPTC and C2PA metadata are some of the technical approaches in use by our organization and by the industry at large today to enable transparency of AI-created or AI-edited content online. This paper examines our organization’s approach to marking AI content and providing user transparency, highlighting lessons learned–and the challenges ahead–in striving for effective AI transparency, including suggestions for research areas most likely to advance industry solutions for indirect disclosure and user transparency for GenAI content. Key challenges have included the lack of robustness of metadata, imperfect robustness of watermarks, the difficulty in defining "materiality" for AI edits, and how to provide users appropriate transparency. We provide details of our experience launching labels for first- and third-party content – both fully AI generated and AI edited – at scale on global apps using GenAI signals from IPTC, C2PA, and known invisible watermarks and the challenge of meeting user expectations related to materiality of edits and choice of language, resulting in changes to our approach.