Meta has introduced Content Seal, an invisible watermarking system designed to flag AI-generated images, following pressure from its own Oversight Board earlier in 2026 to tackle deceptive generative AI content. The technology was announced quietly alongside Meta's new Muse image and video generation tools, but critics argue Meta should have adopted Google's more established SynthID system instead of building its own, especially given the growing risk of AI-generated misinformation.

Content Seal embeds a hidden provenance signal into images that survives cropping, compression, resizing or screenshotting, functioning much like Google's SynthID. However, it currently has significant limitations: watermarks can only be checked via a separate web tool Meta is testing, rather than being integrated into the Meta AI chatbot as Google has done with Gemini, and the system only covers images made with Muse, not content from Meta's older AI tools. A Meta spokesperson said the company is exploring ways to bring detection closer to where users actually encounter AI content, though this was not available at launch.