Does Canva train AI on your designs? The defaults by plan

The Morse teamPublished October 27, 20269 min read

As of October 2026, Canva trains its AI on your designs and uploads only if your privacy settings allow it, and it doesn't for Teams, Business, Enterprise, or Education accounts. For Free and Pro accounts, Canva doesn't publish which way that setting starts, so open Your account, then Privacy controls, and look before your next upload.

We read Canva's Trust Center, privacy policy (updated August 25, 2026), AI Product Terms (effective June 26, 2026) and their archived versions, Terms of Use, Canva Shield pages, privacy settings help article, and its list of subprocessors (updated October 23, 2026) on October 27, 2026. To its credit, Canva publishes a named list of every outside company that runs its AI features, with where each one processes data.

Does Canva use your designs to train AI?

Yes, if your privacy settings allow it, and only on personal accounts. Canva's Trust Center says "Canva won't use your user content to improve AI-powered features unless this is consistent with your Privacy Settings." Its privacy policy says Canva may analyze "your activity, content, media uploads and related data in your account" to "train our algorithms, models and AI products and services using machine learning."

There are two settings. "User content" is "the text in your designs and media you upload." "General usage and activity" is design metadata, "such as how people use layouts, colors and fonts," and analytic events "such as what features you use, and which buttons you click" (Canva). Where each starts, by account type, according to Canva's own pages read on October 27, 2026:

The Affinity row is the closest Canva comes to stating the personal default. Its Canva Shield post, updated in January 2026, says that if you sign up through Affinity Studio, "your content is only used to improve AI-powered features if you choose to opt in" (Canva). We read that carve-out as a sign that ordinary sign-ups start opted in.

Some top results say the opposite. CNET wrote in September 2025 that Canva "automatically opts you out" of training on your content (CNET via Yahoo). Canva's own pages say that only of team, business, school, and Affinity accounts.

The two pages for business accounts also disagree. The newsroom post says team content "cannot be turned on," while the Trust Center says "the ability for Business, Teams, and Enterprise Admins to opt in is coming soon, and we'll let you know when it's ready." If you run a team, watch for that notice.

How do you turn off Canva AI training?

From the Canva homepage, open your Profile icon, choose Your account, then Privacy controls, and use the toggle next to each permission (Canva). The direct link is canva.com/settings/privacy-preferences. Turn off both AI permissions, the one for your content and the one for your general usage. If your profile photo isn't on the homepage, Canva's help page says to look under More, since it's rolling out new navigation gradually.

A lock icon means you can't change that preference, "because you're part of a managed team, or due to your account type." Canva AI's Memories are a separate switch. The privacy policy says Canva may store data from your account "as Memories" and that you manage them under AI personalization settings, not Privacy controls.

What does Canva train its AI on?

Canva's Trust Center names four sources besides your own content (Canva): content from Canva Creators "who have not opted out," licensed content from stock media agencies and partners, openly licensed content, and general usage and activity where your settings allow it. Creators who haven't opted out "receive payment" for it.

Canva's Shield post says it's "committing $200 million in content and AI royalties to be paid to our creator community over the next three years," and that Creators can opt out "when they join the Canva Creator program, or at any time after" (Canva).

In October 2026, Canva released Craft, its first image model built in-house and available to every user in Dream Lab. Canva says Craft's training data "includes millions of human-made templates from Canva's content library," including posters, social posts, presentations, and menus (Canva). If you publish templates as a Canva Creator and never opted out, your work is the kind Craft learned from.

The Creator default has flipped since launch. When Canva introduced Magic Studio in October 2023, Plagiarism Today reported that Creators "will have to opt in" to AI training and "will be opted out by default" (Plagiarism Today). Canva's pages now describe the reverse: Creator content is used unless the creator opts out.

What do Canva's AI Product Terms say, in plain English?

The table puts each clause that touches your work next to what it means in practice. It's our summary, not legal advice, and the full AI Product Terms are short enough to read in one sitting.

The terms changed three times in under a year, by Canva's policy archives:

• Before November 28, 2025, the terms named OpenAI and said that, "subject to your Privacy Settings," third-party providers "may also use your data to improve their services." They also required you to label AI content you posted to social media "in a way no user could reasonably miss." Both are gone from today's version, though terms.law's summary still lists the labeling rule.
• March 16, 2026: The new version said "you do not own your Audio Output" from Canva's AI audio tools, and barred selling it on its own.
• June 26, 2026: The current terms list music, voice, and sound effects as Output and drop the separate audio section, so audio falls under the same ownership line as images.

Which AI companies see what you type into Canva?

Twelve outside companies are listed for "Canva AI Services" on Canva's list of subprocessors, updated October 23, 2026. They include OpenAI, Anthropic, Google, Amazon Web Services, ElevenLabs, Kling AI, Byteplus, and Alibaba Cloud. Four of them, Kling AI, Byteplus, Alibaba Cloud, and Nanonoble, "do not process Customer Personal Data for Canva Education, Canva Enterprise or Customers with a signed Order Form." Even on Canva Education, the privacy policy lists OpenAI as processing student data for "Canva AI products" (Canva).

Canva's pages say what these companies may do with your input, but not how long they keep it. The AI terms limit sharing to "the specific purpose of providing you with the functionality," and the Trust Center says partners may improve AI features with your content only "if this is consistent with your Privacy Settings."

Leonardo.AI, a Canva company since 2024, is on the same list as a Canva group subprocessor (Canva). Leonardo's own app, and the public gallery its free plan publishes to, are covered in our Midjourney alternatives.

What is Canva Shield, and does it stop AI training?

No. Canva Shield is Canva's name for its whole set of AI safety and privacy measures: input and output moderation, a report button, the privacy settings above, the Creator royalties, and indemnification (Canva). The training switch is still the one in Privacy controls.

The legal protection is narrow. Canva says "eligible enterprise customers are indemnified for AI output, at no extra cost, when using our Magic Studio products," for "the rare event of an intellectual property claim." The page offers it to no other plan and doesn't say what makes an Enterprise customer eligible, so ask Canva's sales team before you rely on it.

How Morse keeps your design work private

Canva's AI runs on twelve outside vendors, OpenAI, Anthropic, and Google among them, and each prompt or image you give its AI tools can go to one of them. Whether any of it then improves AI depends on a setting whose starting point Canva doesn't publish for personal accounts.

Morse doesn't train AI on anything you write or make, and it doesn't sell your data or show ads. The images you make in Studio, Morse's image tool, sit in your browser, encrypted, next to your chat history (privacy policy). Clear the browser and they're gone for good, since Morse has no way to restore them, so leave your brand kit and source files where they live now.

Most design work starts as words: a brief, a tagline, a prompt you'll refine. Each chat runs in the privacy mode you pick:

• Private, for the brand voice doc and launch copy you're drafting. New chats start here. Grok and open-weight models such as DeepSeek, Kimi, and GLM answer from hosts that agreed to keep nothing. Where a host confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
• E2EE, a Pro feature, for an unannounced rebrand. Before a word goes out, your browser verifies the enclave's proof that it's genuine and seals the message so only that enclave can open it, so our servers pass along text they can't read. That holds as long as the enclave hardware and attestation work as designed. E2EE has no image generation, so settle the wording there and paste it into Studio.
• Anonymous, for asking GPT, Claude, or Gemini with your identity removed. Morse's proxy takes off your account, email, and IP address and sends the message under Morse's own account. The provider sees it without knowing it's you and may keep it for a limited time to check for abuse, under its own policy, so keep client and product names out of it.

History in every mode stays in your browser, locked when you sign out and opened with two keys, one of which never leaves your device. Because we don't store chats, we usually have very little to give authorities.

Which Canva AI settings to change today

• Free or Pro: Turn off both AI permissions in Privacy controls, and check AI personalization for Memories.
• Canva Creator: If you don't want your templates in Canva's training data, opt out through the Creators support form.
• Teams, Business, or Enterprise: Your content is out of training today; watch for the admin opt-in Canva says is coming, and check the general usage toggle if it isn't locked.
• Drafting the words before the design: Write briefs and prompts in a Morse chat. Morse doesn't train AI on them. For how other apps handle this, see which AI companies train on your data and how Adobe's terms treat your files.

Start a chat from the Morse home page with no account, or sign up free, with no card, for 3 Studio images a day. Pro, at $19 a month, adds E2EE chats.