Photos edited with certain Apple Intelligence tools on iPhone now carry an invisible AI watermark. Starting with iOS 27.2 and the matching 27.2 releases for iPad, Mac and visionOS, Apple embeds Google’s SynthID mark in the pixels of images that those tools create or change. The mark cannot be seen, and nothing in the Photos interface announces it.
AI, explained every day: The Deep View is the go-to daily newsletter for 600,000+ founders, knowledge workers and professionals who want to stay up to date with artificial intelligence.
We may earn a commission if you sign up.
The change was detailed in an Apple support article, and 9to5Mac wrote it up on Nov. 6, less than two weeks after iOS 27.2 reached the public on Oct. 26, according to MacRumors.
Five Apple Intelligence tools that add SynthID
Apple says its watermark is imperceptible to people, does not change how an image looks, and can be read by a SynthID detector. According to Apple’s support document, the tools that add it are Clean Up, Extend and Spatial Reframing in the Photos app, plus Image Playground and Image Wand. Image Playground produces wholly new pictures, while the three Photos tools can alter genuine photos from a person’s own library, which is where the mark matters most.
Pictures made or changed before updating carry no mark unless they are edited again with one of those tools. 9to5Mac’s walkthrough adds that the software covers iOS, iPadOS, macOS and visionOS 27.2.
The split between creating and editing is the detail with the most consequence. A picture generated from a text prompt in Image Playground is plainly synthetic, and a mark on it surprises no one. A snapshot of a family dinner that went through Clean Up or Extend is a real photograph that was altered, and after the edit it carries a machine-readable flag saying AI took part. Apple’s wording is that the watermark can indicate AI involvement, which describes participation rather than authorship, so a detector reading it cannot say how much of the frame was changed.
Uneven results in hands-on testing
The watermark does not appear on every edit. MacRumors tested the feature and reported that every Image Playground image carried a SynthID mark, and that edits made with Extend and Reframe in Photos did as well when checked. Clean Up was less consistent: small edits showed no watermark, while larger ones did. Anyone who removes a stray object from a vacation photo may therefore end up with a clean image, and someone who erases a larger area may not.
Apple has not published a size threshold, and its support article does not explain the difference.
Neither source says whether the mark survives every kind of editing or how it could be removed. Google’s own description says SynthID is designed to hold up through cropping, filters and compression, as MacRumors relays, but no independent test of that claim on Apple’s images appeared in the material reviewed.
Checking an image at synthid.com
The iPhone has no built-in way to read the mark. Detection happens on the web, where an image can be uploaded to the SynthID Detector at synthid.com and checked for the watermark. Google created the technology to identify AI-generated content and embeds it directly into the pixels of images and videos instead of in metadata, so it is not lost when a file’s information fields are stripped on upload to social platforms. Google describes the system in its SynthID overview from Google DeepMind.
The choice of a Google-built mark for Apple’s images is a notable pairing. Apple announced plans to add SynthID support in June, and MacRumors reports that other companies, including OpenAI, use the same technology, which raises the prospect of a shared way to flag machine-made images across rival platforms. A mark only helps if detectors exist where images are viewed, and for now the public check requires visiting a separate website.
For ordinary users the practical effect is modest today. Photos look identical before and after the update, the editing tools work the same way, and no setting appears in the 27.2 interface to switch the mark off, as far as the 9to5Mac description and Apple’s page indicate. The effect shows up later, when a watermarked picture is shared and someone, perhaps a newsroom, a school or a platform trust team, runs it through a detector and gets a positive answer that the sender never saw coming.
Rival phone makers already face the same question about edited images, and a common detector would let one check cover many brands. Whether Apple’s choice of Google’s system becomes the industry norm depends on other companies adopting it, a step the cited reports mention only in passing.
The unresolved piece is who will check. Apple’s support page speaks of AI involvement being indicated, but it does not describe any Apple-side labeling of watermarked images in Photos, Messages or Safari, and it does not say whether such a feature is planned. Until Apple or another party builds that visible layer, the invisible mark remains a tool for people who already know to look for it and who take the extra step of uploading a picture to Google’s checker.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
AI, explained every day: Get The Deep View →
We may earn a commission if you sign up.
More from Morning Overview
- Common allergy, bladder and sleep pills tied to sharply higher dementia odds
- NOAA now gives this winter a 75% chance of the strongest El Nino since 1950
- Doctors warn a silent liver disease now affects one in three American adults
- The FBI told the rest of ShinyHunters to surrender after a 24-year-old was arrested
