iOS 27.2 introduces SynthID to recognize images created or modified with AI

iOS 27.2 SynthID - Apple has introduced with iOS 27.2 something new regarding the recognizability of images processed through artificial intelligence. Image Playground creations and some photographs edited with Apple Intelligence now incorporate SynthID, the digital watermark developed by Google to identify content generated or retouched with AI tools. The support, announced by Apple in June, thus becomes operational in the update. The peculiarity is that the mark does not appear visually on the photograph and does not require any additional operation on the part of the user.

How SynthID watermarking works

iOS 27.2 SynthID - SynthID uses a different identification system than traditional visible labels or information inserted into a file's metadata. The watermark is embedded directly into the image pixels, creating an invisible signal that can be detected through dedicated detection tools.

This choice is important because metadata can be deleted when editing, exporting, or sharing a photograph. A mark integrated into the visual content, however, is designed to be more resistant to subsequent transformations.

According to Google, SynthID does not compromise the perceived quality of the image and is designed to resist common operations such as cropping, filtering and compression. The system can also be used for videos, although the novelty described for iPhone specifically concerns images.

The technology is not reserved for Google products. OpenAI and other companies also use SynthID, contributing to the spread of a shared system for recognizing certain content created with artificial intelligence.

Which images receive SynthID with iOS 27.2

The introduction of the watermark does not mean that any photograph edited on iPhone will automatically be identified as AI content. The behavior depends on the tool used and, in some cases, on the extent of the intervention carried out on the image.

The main distinction is between images generated entirely with Image Playground and photographs processed through specific Apple Intelligence features in the Photos app.

Image Playground and Photos app tools

All images created with Image Playground receive the SynthID watermark. In this case the mark directly accompanies the content produced through Apple's generative tools.

For photographs edited in the Photos app, however, the behavior is more complex. According to the tests reported by iSpazio, the interventions carried out with Extend and with Spatial Reframing produce images containing the watermark.

The Cleanup function has an additional distinction. In the tests described by the newspaper, a small retouch did not result in the inclusion of SynthID, while larger changes produced marked photographs.

Therefore, there is no rule according to which every use of Clean up necessarily involves the application of the watermark. The available documentation describes different results in relation to the extent of the changes, without indicating a precise technical threshold above which the flag is added.

How to check if a photo contains SynthID

The presence of the watermark cannot be recognized simply by looking at a photograph. SynthID is invisible and does not add symbols, writing or other graphics to the image.

Currently, iPhone does not offer a built-in command to directly check the presence of the mark. To verify you need to useSynthID Detector by Google, the tool dedicated to signal recognition.

The procedure consists of uploading the image to the service and submitting it for analysis. The detector allows you to check whether the file contains the signal associated with artificial intelligence tools that use SynthID.

It is important to distinguish the detection of a watermark from the ability to recognize any artificial image. The system identifies its own mark: the absence of a detected signal does not, in itself, prove that a photograph was not generated or modified with other technologies.

What changes for those who create and share images

For those using Image Playground or Apple Intelligence's photo editing tools, the introduction of SynthID brings no noticeable changes to your normal workflow. Images continue to be created and edited without showing additional visible labels.

The difference mainly concerns the possibility of subsequently identifying the intervention of artificial intelligence. Anyone receiving an image can use a compatible tool to search for the mark and get an indication of its origin or processing.

The declared resistance to cropping, filtering and compression takes on particular relevance in online sharing, where platforms can automatically modify uploaded files. However, this is not the same as guaranteeing that the watermark is detectable after any transformation.

According to what was reported by iSpazio, social networks also use or will be able to integrate recognition systems of this type to identify AI content. However, the source does not specify which platforms have already completed the integration nor does it provide a timetable for those still affected in the futureupdates.

Apple's adoption therefore expands the use of a technology already used by other companies. The concrete result is an identification mechanism that accompanies some images generated or modified on the iPhone without visibly altering their appearance, making subsequent verification possible using dedicated tools.

Editorial source consulted:iSpace.