iOS 27.2: Apple's AI images now carry Google's SynthID watermark

Images from Apple's AI now carry an invisible mark. With iOS 27.2, the iPhone tags everything Image Playground creates with SynthID – the watermark developed by Google. Apple announced the move back in June along with the revamped Image Playground, and now it is here. What gets marked, and how can you check it?

For newly generated images, things are clear: they always get the watermark. For edited photos, it depends on how heavily the AI intervenes. MacRumors tried this in the Photos app: images changed with Extend and Reframe carried SynthID, while with Clean Up it depended on the size of the edit – small retouches stayed unmarked, larger ones got the watermark. That is a reasonable line, because removing a piece of lint from a sweater does not make an AI image.

SynthID does not sit in the metadata, which is quickly lost when sharing or taking a screenshot, but directly in the pixels. Google promises that the watermark does not affect image quality and survives cropping, filters and compression.

The watermark cannot be checked on the iPhone itself, though, as there is no built-in detection. For that, you need Google's SynthID Detector, which Google opened up to everyone in early October.

What stands out most is that Apple is not building its own solution here but relying on a competitor's technology. That is pragmatic: a watermark is only useful if as many providers as possible use the same one and it can be checked in one place. With Apple on board, SynthID is a good deal closer to a real industry standard – even if it still only shows what has been marked, not what someone has created with other tools.