Article · AI translation
IT Home reported on November 7 that Apple updated its official support document on November 5, announcing that starting with iOS / iPadOS / macOS 27.2 and later versions, it is introducing invisible SynthID watermarks for some Apple Intelligence image tools.
Citing a blog post, IT Home said the affected image tools include Image Playground and Image Wand, as well as the Clean Up, Extend, and Spatial Reframing features in Photos.
Notably, the marking is not limited to images generated entirely by AI. Even photos actually taken by users will receive a SynthID watermark if they are processed with the specified AI tools above.
Apple said these watermarks are imperceptible to the human eye and do not alter the appearance of an image, but can be read by a SynthID detector. To check whether an image contains a watermark, visit synthid.com in a web browser, upload the image, and follow the on-screen steps to view the result.
At its fall launch event this year, Apple announced “Apple Reference Images” for the iPhone 18 Pro series. By supporting the SynthID standard, the feature supports pixel-level labeling of authentic image data, allowing users to use the technology to prove the authenticity of photos they have taken.
According to Apple's official description, Apple Reference Images start from the most secure point possible: the new main camera securely labels photo data at the pixel level, while dedicated cloud computing converts the labeled data into an immutable image, saved in the Photos app for users to view.
Related reading:
“Proving a photo isn't AI: Apple's iPhone 18 Pro supports pixel-level labeling of authentic image data, but it still has nothing to do with China”
“Apple's iOS 27.2 adds hidden SynthID watermarks to AI-generated images”