What is AI watermark?

An AI watermark is a signal embedded in AI-generated content — visible or invisible — that identifies it as synthetic. Visible watermarks (a logo or label burned into the corner) are the legacy form, mostly replaced by invisible techniques because visible marks reduce commercial usability. Invisible watermarks fall into two families: (1) statistical signals in pixel/audio/token distributions (SynthID, Meta's Imagining, OpenAI's video watermark), and (2) cryptographically signed metadata (C2PA). Major image models — Imagen 3, GPT Image, Flux Pro, Midjourney, Stable Diffusion 3 — now ship invisible watermarking by default. For brands using AI UGC commercially, this matters because platforms (Meta, TikTok, YouTube, LinkedIn, Pinterest) are increasingly scanning uploads for these signals and auto-applying 'Made with AI' labels — so manual disclosure may not be optional even if the brand prefers not to label.