Does Canva add Content Credentials or AI metadata to its images? What two Canva files and Canva's own pages show
The two Canva pages we read do not say Canva adds Content Credentials to its images, so we do not say it either. Canva's AI Product Terms name C2PA metadata when they bar removing provenance tags. In two real files filed under Canva, our checker found no Content Credentials record.
What do Canva's own pages say?
We read two Canva pages in a browser on October 30, 2026. Neither says that Canva writes Content Credentials, C2PA records or AI labels into the images its tools produce. Neither names a Canva product that does. We also found no mention of a visible watermark on either page.
Here is what the pages do say.
| Canva page | What it says | What it does not say |
|---|---|---|
| AI Product Terms (effective June 26, 2026) | Among the things users may not do with Canva's AI products: mislead anyone that AI-made content was made by a person, and remove, alter or disable provenance or metadata tags from AI-generated content, including C2PA metadata. | Which Canva products write such tags, in which file types, or whether a given export keeps them. |
| Student Safety and Trust page for Canva Education | Among the safety measures it lists for AI are plain-language AI terms, in-product transparency and "C2PA membership". | What the membership means in practice, or that images carry a record. |
The terms page names C2PA metadata as an example of a tag users must not remove. That is a rule about users. It is not a statement that every Canva image carries such a tag. Our guide on which AI image generators add watermarks compares what other makers say, and What are Content Credentials? explains the record itself.
What did our checker read in two files filed under Canva?
We hold two real files filed in the Wikimedia Commons category "Images generated by Canva": one PNG and one JPEG. Third parties uploaded them, and we do not know how Canva produced or exported them, or what happened between export and upload. They show what our checker reads in two files we hold. They do not show what Canva does, and we do not call them typical of Canva.
The results below come from our own test files, checked on October 30, 2026. The counts for our whole real-file set are on our accuracy page.
| File | Content Credentials record | Metadata we found | Headline |
|---|---|---|---|
| PNG, 48 by 49 pixels | None | One XMP chunk, with no statement of how the image was made | Metadata found; source not established |
| JPEG | None | EXIF and XMP; the creator tool field names Canva; nothing says how the image was made | Metadata found; source not established |
The PNG is only 48 by 49 pixels. Our Stable Diffusion watermark check needs at least 256 by 256, so it was skipped for that file. For both files the Content Credentials and metadata checks ran. The Stable Diffusion check ran on the JPEG and reported no match.
A tool name in a creator tool field says which program wrote the file. It does not say whether the picture came from an AI tool, a template or a photo. That is why the checker does not treat it as a source declaration, and why neither file reads as declaring AI use.
Is Canva on the trust lists our checker loads?
A Content Credentials record is signed with a certificate. The C2PA Trust List and the frozen Interim Trust List are lists of signers a checker can recognize. We read the subject of every certificate in three of our list files on October 30, 2026: 32 signer anchors on the C2PA Trust List, 27 anchors on the Interim Trust List and 115 certificates on its allowed list, 174 in all. None names Canva.
The lists were fetched on October 27, 2026, and we did not search the 23-certificate time-stamp list. A list holds signers only. A missing name does not show that Canva writes no records, and a present name would not show that a product signs by default. It shows only what those certificates say about themselves. Our guide to the two trust lists says who is on each.
What if a Canva image has no record?
Finding no record means only that our checker found none in that file. Content Credentials can be lost through everyday actions, such as taking a screenshot or converting a file, according to OpenAI's help article on provenance signals. That is OpenAI's statement, in an article about its own content. It shows why an absence has several possible causes, and none of them is visible from the result.
The same holds for the two Canva files. A record may never have been written, or may have been dropped on the way to Commons. We cannot tell which. Our guide to where AI declarations live in an image file shows where the other signals sit, and our guide to Microsoft Image Creator, Paint and Designer covers what Content Credentials those tools add and how a checker reads them.
What this means when you check an image
Open the checker, choose the file, and look for the headline "Metadata found; source not established". The line under it says: "The file carries metadata, but none of it says how the image was made." That is what both of our Canva files showed.
It tells you the file has some metadata, for example a tool name, and that none of it says whether a person or an AI tool made the picture. It does not tell you the image is real or human-made. It does not tell you the image came from Canva or from anywhere else. The checker's limits line applies: these checks cannot determine whether an image is AI-generated.
If you need to know what Canva puts in its own exports, the useful source is Canva. The terms page above has a link to Canva's Trust Centre. Check the original downloaded file where you can, because a screenshot or a converted copy can lose its signals. Our Privacy and method page explains what our checker does with your file.
A record, if a file did carry one, would read differently. How to check Content Credentials walks through each result.
Guides on this topic
- How to tell if an image is AI is the main page on this topic.
- Which AI image generators add watermarks? compares what each maker says it adds.
- Microsoft Image Creator, Paint and Designer covers what Content Credentials those tools add and how a checker reads them.
- Where AI declarations live in a file maps XMP, EXIF, IPTC, PNG text and C2PA.
- How to check Content Credentials explains each result step by step.
Frequently asked questions
- Does Canva add Content Credentials to its images?
- We cannot say. The two Canva pages we read name C2PA metadata in a rule for users, and list "C2PA membership" under transparency, but neither says Canva adds a record to its images.
- Does Canva put a visible watermark on AI images?
- We found no mention of one on the two pages we read. Our guide to visible marks lists the marks we could source.
- Do our two Canva files show what Canva does?
- No. Third parties uploaded them to Wikimedia Commons, and we do not know how they were exported or handled. They show two files, not a product.
- Does our checker read any hidden Canva tag?
- The checker has no Canva-specific check, and the two pages we read document no Canva tag. A tool name in a creator tool field is read as ordinary metadata. The checker reads Content Credentials, XMP, EXIF, IPTC, PNG text and the Stable Diffusion watermark.