AI watermark (marking AI-made content: SynthID, C2PA and more)
Japanese AIウォーターマーク
An AI watermark is a mark left in images, video or audio made by AI, saying that AI made it. It helps people and tools check where content came from later.
There are two main kinds of AI watermark. One is an invisible mark hidden inside an image or sound, such as Google's SynthID. The other is history information attached to a file, such as Content Credentials from the C2PA standard. The first is like a faint pattern in paper, and the second is like an ingredient label on a product.
Watch out These marks only say that AI made something. They are not permission to use it freely. Also, no mark does not prove that a person made it. When you use AI output for work, check the service's terms as well.
Source: Sources: Google DeepMind, "SynthID"; C2PA; Content Authenticity Initiative