Google Messages adds AI-image provenance checks in testing

Google Messages for Android appears to be preparing a new “View details” option that can flag whether an image is AI-generated or edited with AI tools, using C2PA Content Credentials-style analysis. The feature isn’t live yet, but app code strings already outl

The next time someone sends you an image in Google Messages, you may soon be able to tap for more than just what it shows. The app looks like it’s moving toward a way to tell, with more nuance, whether the picture is wholly AI-made or whether AI was only used for edits.

In an on-device update being tested by developers. Google Messages appears to be working on analysis for C2PA Content Credentials. a system designed to attach provenance information to media. The goal isn’t framed as a simple yes-or-no verdict; the wording in the app suggests it will distinguish different kinds of AI involvement—down to cases where images were captured by a camera and cases where multiple pieces of media were combined.

Today’s look is tied to Google Messages for Android in a specific test build: version messages.android_20260611_04_RC00.phone.openbeta_dynamic. Even though the tool hasn’t been visible in action yet, text strings found inside the app point to what’s coming.

One of the strings references viewing information tied to an image—suggesting this will be accessible through the “View details” option under the overflow menu after tapping an image shared in a chat. When people select that option. the app is preparing to show descriptions about what kind of editing or creation may have happened.

The menu language is already granular. It includes labels such as “Edited with AI tools,” “Edited with multiple AI tools,” and “Media made with AI.” It also lists versions that split out non-AI editing (“Edited with multiple non-AI tools,” “Edited with non-AI tools,” “Media made with non-AI tools”).

Other strings show Google Messages is also preparing to handle mixed-origin media. There are options that suggest the app will report when only parts of an image may include AI content—phrases like “Parts of this media were made with AI” and “Parts of this media may have been made with AI.” There are also statements aimed at screenshots and recordings. including “Parts of this screenshot or recording may include AI content.”.

The most telling detail is that the system appears designed to classify not just the presence of AI. but the production path. The code strings include options such as “Media captured with a camera without software adjustments. ” “Media captured with a camera. ” and “Media captured with a camera. multiple images were combined.” There are also combinations-based entries like “Multiple pieces of media combined. ” “Multiple pieces of media combined. some may have been made with AI. ” and “Multiple pieces of media made with AI combined.”.

All of it connects to a bigger effort underway across Google’s products: making provenance checks less manual. Google has previously referenced SynthID. a system that can help scan media for AI content. and it was also baked into Gemini for easier access across devices. Separately. the C2PA Content Credentials system—another provenance approach—has been part of the broader conversation about how to verify content origins in an era where images can be generated or edited at scale.

In Messages, the feature still isn’t functional for users—at least not in what’s currently visible. But the groundwork is clearly there. The question now isn’t whether a tool like this is possible. It’s whether people will actually use it once it’s one tap away in the chat UI. instead of having to hunt down verification tools elsewhere.

For now, Google Messages is still in the build-and-test phase. But the wording already shows the company is preparing for a messy reality: not every image is fully AI-generated. and not every “edited” photo should be treated the same. If this ships as the code strings suggest, the inbox could soon come with receipts.

Google Messages Android AI imagery C2PA Content Credentials AI detection provenance SynthID Gemini