The problem with AI-created images is that a simple instruction can produce a relatively realistic image of anything. It is not difficult to imagine what problems can arise.
Google introduced a feature called Synth ID The aim is to combat fake images and the misuse of AI-generated images.
According to the company, this technology“embeds a digital watermark directly into the pixels of an image, making it imperceptible to the human eye but recognizable for identification.”
To the naked eye, the image generated by the AI appears normal. But it will have an easily recognizable mark so people will know it’s not an authentic image.
Currently, SynthID is available to a “limited number” of customers using Image, Google Deepmind’s model that uses text instructions to create photorealistic images.
According to Google, SynthID is said to be more effective than traditional watermarks, which are more easily removed, truncated, or unsightly on an image. The SynthID watermark is preserved even if you add filters, change colors, or resize the image.
The SynthID tool should be able to tell users whether Image was used to create an image using a system of three levels of confidence: probable, unlikely, and possibly detected.
Going forward, Google has indicated that it could expand the tool to be able to identify images of other models and integrate the SynthID tool into other products it offers.