Google Cloud, in partnership with Google DeepMind and Google Research, has launched a new tool, SynthID to help identify artificial intelligence (AI)-generated images. SynthID puts a special code called a watermark into these images, but it is hidden so well that humans cannot see it. However, computers can detect this watermark, making it easier to spot these images.
Traditionally, watermarks were used to mark images, but they were easy to remove or change. SynthID’s watermark is different because it does not hurt the image’s quality. Even if the image’s colours or size is changed, the watermark stays hidden, and computers can still find it. SynthID uses two smart systems to do its job: one for adding the watermark and one for finding it.
The new technology is being made available to some special customers who use Vertex AI and a model called Imagen, a smart system that turns text into life-like pictures.
The AI-generated images that are becoming more and more realistic are created by computers and can sometimes be mistaken for real ones. The technology behind these AI-generated images, Generative AI is advancing quickly. While AI can be used for creative purposes, it can also be used to create fake images that spread false information.
