Does Krea train on your data?
Krea's Business terms promise not to train on your inputs or outputs. The terms and privacy policy for Free, Basic, Pro and Max never say whether yours are trained on.
Graded on the plan most people are on. The other plans are in the matrix below.
This clause is real, but it doesn't resolve the training question - see the finding below.
"When you submit User Submissions through the Services, you hereby do and shall grant Krea a worldwide, non-exclusive, perpetual, royalty-free, fully paid, sublicensable and transferable license to use, edit, modify, truncate, aggregate, reproduce, distribute, prepare derivative works of, display, perform, and otherwise fully exploit such User Submissions"
- 2 Oct 2026
- 5 Oct 2026
- Free, Basic, Pro and Max
- global
- Fetched from the company
- 5C12608A…DC8B
Why this is still UNCLEAR: Krea's Terms of Use and privacy policy, both last updated 20 May 2024, never use the word train. The terms take a perpetual, sublicensable licence to "otherwise fully exploit" whatever you submit "in connection with this site, the Services and our (and our successors' and assigns') businesses". The privacy policy lists "Improving the Services, including testing, research, internal analytics and product development." Neither says whether that includes training Krea's models. Krea does make the promise, in its Business Terms of 1 January 2026, for Business plans only. For everyone else the documents leave the question open.
"Krea commits that it will NOT use your Inputs or Outputs to train its AI models or algorithms."
- 2 Oct 2026
- 5 Oct 2026
- Business
- global
- Fetched from the company
- 2AD6A420…B679
How we read it: Krea's Business Terms, last updated 1 January 2026, govern the Business plan for team workspaces. They commit that Krea "will NOT use your Inputs or Outputs to train its AI models or algorithms", and repeat it for team accounts and for content made while a Business subscription ran. Inputs are defined as text prompts, images, reference materials or other data you provide. B and not A because it's a contractual promise. Nothing says Krea can't reach the content.
What Krea said when we showed them this
Clarified“Regarding whether Krea uses customer data to train AI models, Krea states it does not. For Business and Enterprise plans, Krea commits that Inputs and Outputs are not used to train its AI models. [...] If you were looking specifically at the general Terms of Use and Privacy Policy, those pages do not spell this out. The training policy is published in the Business Terms and the Enterprise page above.”
What each plan can refuse
The same questions, asked of every plan we could identify. Every cell comes from a quoted clause or a documented search - see the evidence above.
- Not established
- Not stated
- —
- —
- Not established
- Nothing to switch off
- Not stated
- 20 / 100
- Not stated
- Unknown
How to opt out of Krea AI training
Plan by plan, from what Krea's own policy says about refusing. Where it publishes the steps, they're listed here with the limits they carry. Where there's nothing to switch off, or nothing graded, this says so instead of inventing a menu path.
Free, Basic, Pro and Max
Not graded, so no route is pointed at. Krea's Terms of Use and privacy policy, both last updated 20 May 2024, never use the word train. The terms take a perpetual, sublicensable licence to "otherwise fully exploit" whatever you submit "in connection with this site, the Services and our (and our successors' and assigns') businesses". The privacy policy lists "Improving the Services, including testing, research, internal analytics and product development." Neither says whether that includes training Krea's models. Krea does make the promise, in its Business Terms of 1 January 2026, for Business plans only. For everyone else the documents leave the question open.
Business
Nothing to switch off: Krea's policy says training is off here.
What this policy has done since we started watching
The record starts here. Krea's 3 documents were first captured on 2 Oct 2026, and there's no earlier copy to compare them against, so this verdict has no history behind it yet. From now on every re-read is diffed against those copies, and anything that moves appears here.