Gemini 2.5 Flash Image

Very impressive.

I have to say while I'm deeply impressed by these text to image models, there's a part of me that's also wary of their impact. Just look at the comments beneath the average Facebook post.

I have been testing google's SynthID for images and while it isn't perfect, it is very good, insofar that I felt some relief from that same creeping dread over what these images will do to perceived reality.

It survives a lot of transformation like compression, cropping, and resizing. It even survives over alterations like color filtering and overpainting.

facebook isn't going to implement detection though. Many (if not most) of the viral pictures are AI-generated. and facebook is incentivized to let their users get fooled to generate endless scrolling