iOS 27.2 adds SynthID watermarking to Photos and Image Playground

Apple's release notes for iOS 27.2 list support for SynthID watermarking, which is meant to help show whether a picture in Photos or Image Playground was made or changed with generative tools. 9to5Mac reports that the iPhone update arrived on Oct 26, 2026, and that SynthID support is one of several new features in it.

SynthID is Google's technology, not Apple's. Apple's choice to use it links iPhone pictures to a system that other AI companies also support. The move means pictures made or edited with some Apple Intelligence tools can carry a mark that software can read but people cannot see.

Key facts

• At its WWDC event in June, Apple said photos changed with Apple Intelligence would get a hidden SynthID watermark automatically.
• Apple also said images from the new Image Playground, which can now make photorealistic pictures with a model running on Private Cloud Compute, get the same hidden mark.
• The Verge reports the watermark covers edits made with three Photos tools: Extend, Clean Up and Spatial Reframing.
• Ars Technica explains that SynthID is a signal tucked into the pixels of AI pictures and video, and into the sound wave of AI audio.
• Google opened a public detector at SynthID.com on Oct 7, 2026, that reads marks from partners including OpenAI, Nvidia and Kakao, with Apple listed as coming soon.

Background

Before it switched to SynthID, Apple marked the file data of Image Playground creations and Clean Up edits using a forensics feature of its own, The Verge reports. Metadata is a label attached to a file, while SynthID sits inside the image itself.

Image Playground used to stick to cartoon-style art, and The Verge says Apple had stayed away from photorealistic AI images until this update. When Apple launched its Clean Up tool for removing objects two years ago, software chief Craig Federighi said Apple wanted to "purvey accurate information, not fantasy," The Verge recalls.

The Verge also notes that SynthID has been spreading across the tech industry and that OpenAI recently adopted it. Ars Technica reports Google's claim that Gemini alone has put SynthID on more than 180 billion images and videos, along with audio adding up to 240,000 years.

What it means for you

For most iPhone owners, nothing changes on screen, and no extra setting is needed. Apple says the watermark on generated images and AI-adjusted photos is hidden and added automatically. The goal is to give others a way to tell later whether AI was involved in a picture.

The detector website is free, but Engadget reports that users must log in using an Apple, Google or OpenAI account. According to Ars Technica, Google says each user gets a daily allowance of about 10 checks across images, video and audio.

There are limits. Engadget notes the detector cannot tell content made entirely by AI apart from content that was only edited with it. Ars Technica says some AI companies, such as Meta, use other watermark systems, and open-weight models may add no mark at all, so SynthID will miss some AI content.

Ars Technica says all AI output from Google's Gemini models includes a hidden SynthID tag, so it is difficult to present as genuine. Apple's adoption adds another large source of images to that system. For people who share photos on social media, the change means a shot changed with AI on an iPhone may later be flagged as edited, even if the change was small.

It is not yet clear whether the public detector reads Apple's marks. Engadget reported that at launch, Google listed support for Apple Intelligence images, both generated and edited, as coming soon.

Practical steps

• Install iOS 27.2 by going to Settings, then General, then Software Update.
• To check a file, upload it at synthid.com, which Engadget says anyone can use.
• Log in to the detector using an Apple, Google or OpenAI account, as Engadget describes.
• Plan checks with care, since Google sets a daily limit of about 10 per user, according to Ars Technica.
• Treat a result with no watermark as a lack of evidence, not proof that a picture is real.

What happens next

Google says it is "continuously working on extending our partnerships to allow detecting more generated media," Engadget reports. Engadget also says ElevenLabs has pledged to add SynthID support.

Apple has not said whether it will build a way to check SynthID marks into the iPhone.

FAQ

Can people see the watermark? No. Apple says the SynthID watermark on generated images and photos adjusted with AI is hidden.

What Photos edits get the mark? The Verge names Extend, Spatial Reframing and Clean Up.

If no mark turns up, is the picture authentic? No. Ars Technica says a lot of AI content will slip past the SynthID tool, and the site carries a disclaimer saying so.

Does it cost anything to check a picture? No, the detector is free, though Engadget says an Apple, Google or OpenAI login is required.

Which other companies' marks can the detector read? Ars Technica says the detector supports marks from partners including OpenAI, Nvidia and Kakao.

What is new about Image Playground? Apple says it can now produce photorealistic images using a new model hosted on Private Cloud Compute. The Verge says the tool used to stick to cartoon-like styles.

When did Apple first announce the plan? Apple first described the SynthID plan at WWDC in June.

Do the new Siri AI features work on every iPhone? No. Apple's release notes say they need an iPhone 16 model or later, or an iPhone 15 Pro or Pro Max.