// Privacy

Privacy Policy

Effective September 25, 2026

Who we are

Mosa is operated by Ambassadors Lab. This policy describes how we handle information you share with us through this website and the Mosa service. Questions about it can be sent to hello@getmosa.ai .

What we collect

Your account

Anyone can create a Mosa account. When you do, we store your email address, your name and, if your sign-in method provides one, a profile picture. You sign up either with an email address and password — the password is held by our sign-in service and never seen by the Mosa application — or with Sign in with Google. We record whether your email address has been verified, and check that it is not a known disposable mailbox.

Your workspace and content

Everything you create in Mosa is stored so the product can work: workspace and project names, the people you invite, your workflows, the files you upload (images, video, audio and documents), the media and text your workflows generate, run history, and the prompts and conversations you have with the in-product agent. We also record what each run cost, so we can bill it and show you your usage, and we keep a record when you confirm rights and likeness warranties for uploaded media. If you signed up through a campaign or referral link, we store the campaign and referral fields that came with that link on your workspace — see Cookies & analytics.

Payments and trial verification

To start the free trial you verify a payment method through Stripe; nothing is charged for a card, and a bank verification takes and refunds 0.01 EUR. Stripe returns, and we keep, the type of method, its last four digits, for a bank verification the account holder's name as the bank confirms it, and a fingerprint Stripe derives from the card or account number. The fingerprint does not reveal the number; we use it only to make sure each payment method starts one free trial. If you buy budget or a paid plan, we keep the resulting billing records.

Technical data

Our servers log each request to the product, including your IP address, to keep the service secure and to investigate faults and abuse.

On this website, if — and only if — you accept the cookie banner, we also place an advertising cookie that tells us which Google ad brought you to the site. We don't build a profile of you or sell what it records. See Cookies & analytics below.

Closed beta waiting list

Before signup opened, Mosa was available by invitation through a waiting list. If you joined it, we stored:

The email address you provided.
The role you selected and, if you chose to share it, a short note about your biggest AI bottleneck.
Whether you opted in to our newsletter, and an optional referral code if you entered one.
The IP address and user agent of the request, used only to detect abuse.
A timestamp of the submission.

We may also have invited you to complete a Google Form to register your interest in more detail. If you filled it in, we collected your full name, email address, company and role. We used this information only to send you a confirmation email and to invite you to the beta.

How we use it

To provide Mosa: sign you in, store your work, run the workflows and agent requests you start, and share your work with the people you invite.
To run the AI features you choose, by sending the content they need to the provider of the model you picked, as described under Who processes it for us.
To bill you: track usage against your budget and plan, and process payments through Stripe.
To prevent abuse: verify email addresses and payment methods, limit each person and payment method to one free trial, and investigate suspicious activity.
To talk to you about your account: service emails, invitations and billing notices.

We do not sell, rent, or share your information with third parties for marketing purposes. Product-side processing is described further in the Terms of Service.

Google user data

Mosa uses your Google account in two separate, optional ways. Each asks for your consent on Google's own screen before we receive anything.

Sign in with Google. We receive your name, email address, profile picture and Google account identifier. We use them only to create your Mosa account, sign you in, and show who you are to people in your workspace.
Connecting Google Drive and Sheets. If you connect a Google account in Settings, we receive the account's email address and access tokens that let Mosa read and write the Drive files and spreadsheets you choose for it in Google's file picker, plus the files Mosa creates in your Drive. Mosa cannot see the rest of your Drive. We access these files only when you pick them, when you open a workflow that shows them, or when a workflow you run, built by you or by the in-product agent, runs a Drive or Sheets step — to read the files it names or to write its results back. Your Google connection is personal: other people in your workspace cannot use it. When they open a workflow that shows a Drive file, Mosa uses their own Google connection, so they only see files they can already open themselves.

Storage and protection. Access tokens are encrypted before we store them. Content a workflow reads from Drive or Sheets is treated like any other workflow data: it is kept with that workflow's results in your workspace until you delete them. To show a preview of a Drive image, video or audio file on the canvas, Mosa keeps a temporary copy of it, which is deleted automatically 7 days after it was made.

Sharing. We share Google user data only to provide the features you use: if your workflow sends content from Drive or Sheets to an AI model, it goes to the provider of the model you picked, as described under Who processes it for us. We do not sell Google user data, use it for advertising, or use it to develop, improve or train generalised AI or machine-learning models. No one at Ambassadors Lab reads it unless you ask us to, it is needed for security or abuse investigation, or the law requires it.

Revoking and deleting. Removing a Google connection in Settings deletes its stored tokens. You can also revoke Mosa's access at any time from your Google account permissions. To delete your Mosa account and the Google data tied to it, email hello@getmosa.ai .

Mosa's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.

Where it lives

Your information is controlled by Ambassadors Lab. Our servers and database are hosted in the EU. Access is limited to staff who need it to run the service. Some AI providers process content outside the EU; where we know it, the location is given under Who processes it for us.

Who processes it for us

We rely on the service providers (sub-processors) below. Each one receives only what it needs for the purpose described, and only for the purposes set out in this policy. Scaleway, Stripe and the Google services listed under Marketing site and closed beta handle waiting-list data on our behalf under data-processing agreements.

Hosting and infrastructure

Scaleway — EU cloud provider (data centres in Amsterdam) that hosts the servers the site and the product run on, holds our encrypted configuration secrets, and delivers the confirmation and beta-update emails we send you.
Google Cloud — hosts the database that holds your account and waiting-list records, and the object storage that holds every file the product touches: the media you upload, the media your workflows generate, and the previews, thumbnails and proxies we derive from them.
Google Fonts — serves the typefaces the site and the product use. Your browser fetches them from Google on every page load, so Google receives your IP address and browser user agent. This happens whatever you choose in the cookie banner, because no cookie is involved.
Iconify — serves an icon from its CDN in the rare case one is missing from the set we ship with the product. When that happens your browser sends Iconify the icon name, your IP address and your user agent, and nothing else.

Sign-in runs on our own Zitadel installation, so the account record it keeps — your email address, name and profile picture — stays on our infrastructure. If you choose Sign in with Google, Google confirms your identity and shares those details with us; see Google user data below.

AI model and media providers

When you run a workflow node or ask the in-product agent to do something, we send the content that node needs to the provider behind the model you picked. A provider only ever receives your content if you choose one of its models — but you do not have to connect an account for that to happen, because we supply the API keys unless your workspace has added its own. Depending on the node, what we send can include your prompt and system prompt, files attached to the node (images, audio, video, PDF, text, CSV, JSON and XML), and the output of earlier nodes in the same workflow.

OpenAI — GPT text models, GPT Image, and Whisper transcription. Receives your prompts, attached files, and audio you transcribe. We send these requests with OpenAI's response storage switched off.
Anthropic — Claude text models. Receives your prompts and attached files.
Google (Vertex AI) — Gemini text and image models, Imagen, and Veo video. Receives your prompts and attached files. Two things to know: Veo runs in a United States region because Google offers it nowhere else, and before certain BytePlus video generations we send a downscaled still of your reference image to Gemini to check automatically whether it shows a real person.
BytePlus — Seed text models, Seedream images and Seedance video, plus the DeepSeek and Kimi models BytePlus hosts. Receives your prompts and reference media. We call its Asia-Pacific (Southeast) endpoints. Where a reference image shows a real person, that image is uploaded to a private asset library in BytePlus's ap-southeast-1 region first, because the model rejects such images sent inline.
xAI — Grok text models. Receives your prompts and attached images.
Fal.ai — runs a large part of our image, video, audio, upscaling and transcription catalogue, including models it hosts on behalf of others (Kling, Flux, Recraft, Ideogram, Topaz, GPT Image and Grok Imagine). Input files are uploaded to Fal's own CDN storage before the model runs, so your media is stored there as well as passed through.
Kling AI — image and video generation when you pick a Kling model directly rather than through Fal. Receives your prompts and reference media. We call its Singapore endpoint.
Luma — image generation and editing and Ray video. Receives your prompts and reference media.
ElevenLabs — text-to-speech, speech-to-speech and transcription. Receives the text you have spoken, or the audio you convert or transcribe. If you use voice cloning, your reference recording is uploaded and a temporary voice is created on our ElevenLabs account and deleted again afterwards.
Magnific (Freepik) — image and video upscaling. Receives the media you upscale.
Serper — powers the Image Search node by running your query against Google Images. Receives the search term you enter, and nothing else.
Vocoda — an additional speech-synthesis option, which unlike the providers above runs on Ambassadors' own infrastructure. Receives the text you have spoken and, for voice cloning, your reference audio.

Services you connect yourself

These are reached only if you connect the account in Settings. Until you do, nothing is sent to them.

Google (Drive and Sheets) — if you authorise the Google connection, the Drive and Sheets nodes read files and spreadsheets from your account and write results back to it. The authorisation only covers the files you pick for Mosa and the files Mosa creates, not the rest of your Drive. You can revoke it at any time in your Google account.
Figma — if you add a Figma access token, the Read Figma node fetches the structure of the file you name and the images Figma exports from it.
Cube — if you connect a Cube account, workflows can read assets from your Cube library and write generated media back to it.

Payments

Stripe — our payment processor. Signing up verifies a payment method through Stripe: a card, or a bank account via iDEAL, Wero or Bancontact. Your card or bank details go to Stripe directly and we never see or store them; we keep only what Stripe returns to identify the method later — its type, the last four digits, and for a bank verification the account holder's name as the bank confirms it. Stripe also processes any budget you buy and any paid plan. To do that we give Stripe your email address, your name, and your workspace name, and Stripe collects your billing address and any VAT number from you directly so it can work out the tax.

Marketing site, closed beta, and sign-up

Google (Analytics) — measures how the site is used, only if you accept cookies. If you decline, no usage data is sent to Google. This may involve a transfer of usage data to the United States under the EU–US Data Privacy Framework.
Google (Ads) — measures which of our ads led to a visit or a completed sign-up, only if you accept cookies. If you decline, the tag is never loaded. This may involve a transfer of usage data to the United States under the EU–US Data Privacy Framework.
Both of the above, in the app — sign-up finishes in the app rather than on this site, so the same two Google tags load at app.getmosa.ai at the moment a sign-up completes, and only if you accepted cookies here. What we send is that a sign-up happened: not your name, not your email address, and not the address of any workspace, project, or workflow you go on to open. The tags are not loaded on any other page of the app, and once the sign-up has been recorded we switch them off for the rest of that browsing session, so nothing you do in the product afterwards is measured. If you declined, or never saw the banner, neither tag loads in the app at all.
Google (Forms) — hosted the closed beta interest-registration form through which we collected your full name, email address, company, and role. Responses are stored in Google Workspace and may involve a transfer of data to the United States under the EU–US Data Privacy Framework.

Cookies & analytics

Neither this site nor the app loads analytics or advertising scripts by default. On your first visit we show a cookie banner asking whether you accept them:

If you accept, we load Google's tag script, which sets cookies to measure site usage (Google Analytics, with IP anonymisation enabled) and to attribute a visit or sign-up to the Google ad it came from (Google Ads). The second of those is an advertising cookie.
If you decline or ignore the banner, the script is never loaded and neither cookie is set.

Your choice — whichever way it goes — is remembered in your browser's local storage and in one cookie on getmosa.ai that records the choice itself. That cookie is how the app at app.getmosa.ai knows what you decided, so that a decline made here is honoured there too and you are not asked twice. It carries nothing but the word granted or denied, it is not an analytics or advertising cookie, and it expires after six months.

Separately, if you arrive with a campaign or referral link (utm_* parameters or a ref code), we may set a short first-party cookie named mosa_attr on getmosa.ai. It remembers how you found us long enough to survive the sign-up hop into the app, then we store those fields on your workspace so we can tell which campaigns and partners brought sign-ups. It is not shared with advertisers, it does not load any third-party script, and it expires after ninety days. Declining the analytics banner does not clear it — it is not an advertising cookie.

You can change your choice at any time from the Cookies link at the bottom of any page, which brings the banner back. Withdrawing is as easy as accepting. Cookies already set stay in your browser until you clear them, but nothing further is sent once you decline.

Your rights

You can ask us to access, correct, or delete the information we hold about you at any time by emailing hello@getmosa.ai . If you're in the EU/EEA or UK, you also have the right to lodge a complaint with your local data protection authority.

Retention

We keep your account and workspace content for as long as your account exists. You can ask us to delete your account and everything tied to it by emailing hello@getmosa.ai . Billing records we are legally required to keep for tax purposes are kept for the period the law requires.

Closed beta waiting-list entries are kept until the beta program ends or until you ask us to delete your record — whichever comes first.

Changes

We may update this policy as the product evolves. The effective date at the top reflects the most recent revision.

← Back to Mosa