Today, Google DeepMind and Google Cloud launched SynthID, which is a tool for watermarking and identifying AI-generated images.
The watermark is indiscernible to the naked human eye but can apparently withstand common image manipulation like crops and color adjustments. It isn't foolproof, as stated by Google themselves, but it's "a promising technical approach."
"While generative AI can unlock huge creative potential, it also presents new risks, like enabling creators to spread false information — both intentionally or unintentionally. Being able to identify AI-generated content is critical to empowering people with knowledge of when they’re interacting with generated media, and for helping prevent the spread of misinformation."
Today, in partnership with Google Cloud, we’re beta launching SynthID, a new tool for watermarking and identifying AI-generated images. It’s
My initial thought when I saw this was that in order for this to work, not only do people need to actively choose to watermark their AI images (with Google's tool), it also seems like others who want to find out whether an image has been watermarked need to use the specific, no doubt proprietary technology developed by Google. If this is going to be a useful tool in the long run, there needs to be a standard for what technology generative tools use, and that standard tech then needs to be forcefully applied to all AI-generated imagery regardless of which tool is used - otherwise it's neither practical nor trustworthy. (How are we supposed to trust that an image isn't AI-generated if watermarking it as such is voluntary?)
It's the age-old problem: In order for this to have any effect, it needs to be widely implemented; but it can't be widely implemented if it doesn't exist in the first place, and I suppose we have to start somewhere.
