Your work, protected
Your thinking stays yours.
You're handing Bampot your best ideas before anyone else has seen them. That trust is the whole relationship, so here's exactly how we treat your work, in plain terms.
You're handing Bampot your best ideas before anyone else has seen them. That trust is the whole relationship, so here's exactly how we treat your work, in plain terms.
Everything you bring (briefs, context, decisions) and everything Bampot helps you make is yours. You own the output; we own the method. Nothing you create belongs to us.
Bampot keeps a full record of how every idea was reached, first question to final brief. You can defend the thinking in any room, and build it into your own IP.
The output carries a second kind of proof, too. Text Bampot generates holds an imperceptible watermark, applied by the AI provider under the EU AI Act’s transparency rules. It contains nothing about you, your organisation or your work, does not change the output, and does not affect your ownership of it. So if you are ever asked to show the work was AI-assisted and properly governed, you have two answers rather than one: the full record of how it was reached, and a mark on the words themselves.
Nothing from your work is used to inform anyone else's. Each project stands on its own, walled off from every other.
We don't use your written work to train our own models. For text, we work under paid API and Vertex agreements with the AI providers, under which they commit not to use what we send them to train their models, and we only ever use the paid tiers, never a free one. We hold those provider relationships, not you.
Export everything you've made, whenever you want it. Ask us to remove your work, and we will.
More teams now run a new tool past an AI or data council before they can use it. Here is everything that review needs, in plain terms.
Is our data used to train AI models?
No, not by us, and for text not by the providers either: the language models that do the thinking aren't trained on your written work, under our paid commercial agreements with them, and we only ever use their paid tiers, never a free one. Image generation splits by how the render is made: for faithful pack renders we use Google Vertex AI, under agreements where Google commits not to use what we send to train its models; for mood visuals and other pack work we use OpenAI, where we make no such commitment and are candid about it in the downloadable note below.
Could our work ever reach another customer?
No. Every brief runs in isolation: each AI call carries only your brief, with no shared session across users, and every request for your work is checked against your account on our servers before anything is returned.
Who owns the inputs and the outputs?
You do. You own everything you bring and everything Bampot helps you make. We own the method, never your work.
Which providers process our data?
For the reasoning that builds your brief, Anthropic (Claude) by default, with OpenAI (GPT) and Google selectable. Image generation is a stated exception, routed by how faithful the render must be rather than by preference: faithful pack renders to Google Vertex AI, mood visuals and looser pack work to OpenAI. Each is a named sub-processor under a data-processing agreement.
Where is our data held?
In the United States. Bampot is available in the US, UK and other approved regions today; it is not currently available in the EU.
Can we export or delete our data?
Yes, export any time, and request deletion at any point. Closing your account ends access and removes your data, subject to legal retention.
How do we know an output is sound, not a black box?
Every Bampot Brief carries a Provenance Map: each claim and step traces back to a named source, so your team can see exactly how a conclusion was reached.
Is it secure?
Encrypted in transit and at rest, access behind authentication, and every request checked server-side against your account, with no direct database access from a browser.
What about a DPA and certifications?
A data-processing agreement is available on request. Terms and a Privacy Policy are published; our certification roadmap and full security detail are available for review.
Need this in your own paper, or a signed DPA? Get in touch and we will move at your council’s pace.
Need this to file with your review? Download the pack for your AI or data council, and the note on the information you share. Both answer the questions a review asks, in a document you can attach.
Claude is a trademark of Anthropic, GPT of OpenAI, and Gemini of Google. Bampot is independent (not affiliated with, endorsed by, or sponsored by Anthropic, OpenAI, or Google) and uses their models under their commercial terms.