Privacy notice
Accillion AS · Last updated 17 November 2026
The short version. Lippa's core workspace is hosted on EU infrastructure by default. Some features, including selected model providers, dictation and video generation, may process data outside the EEA and are identified in the product or in this notice. We do not use your content to train Lippa models, we do not sell your data, and we use provider API terms and settings designed so that your Lippa content is not used to train third-party models. You can export your workspace data and delete your account, subject to limited retention for security, abuse prevention, accounting, backups and legal obligations. We collect what Lippa needs to run, protect and improve the service.
1. Who is responsible
Accillion AS, a company registered in Norway, is the data controller for Lippa and for this website. Contact: hello@lippa.ai. The Norwegian Data Protection Authority (Datatilsynet) is our supervisory authority.
Our role. For individual users, Accillion AS is generally the controller for account, product, security, billing-support, safety and analytics processing. For future organisation workspaces, we may act as processor for workspace content under a separate data processing agreement, while remaining controller for account administration, billing, security, abuse prevention and legal compliance. If you publish a generated website or shared content, you are responsible for what you publish; Lippa remains responsible for operating the hosting, security and notice-and-action processes. Business customers that need a data processing agreement can request one at hello@lippa.ai.
2. What we collect
- Account data: email address, password hash or Google sign-in identifier, plan and settings.
- Your content: conversations, uploaded files, memory items, artifacts, generated media, and imports you choose to bring in via the browser extension.
- Device and technical data: IP address, device and browser information, session identifiers, security logs, crash and error reports, and approximate location derived from IP where needed for security, fraud prevention, localisation or compliance.
- AI routing and safety metadata: selected model and provider, request timestamps, token and credit usage, safety classifications, blocked requests, abuse reports, moderation decisions, and content hashes (fingerprints) of reported or blocked material.
- Publishing data: shared links, generated websites, custom domains, page assets, public URLs and publication settings.
- Age and eligibility data: age confirmations and feature-access status where required for media generation or other restricted features.
- Billing data: payments are processed by Stripe. We never see or store your full card details.
- Correspondence: support and contact emails, and mailing-list signups (email plus the interests you tick).
- Voice, when you dictate. If you use the microphone to dictate, we stream your audio to be turned into text. We don't store the audio. The text goes into your message box. We don't keep it on our servers unless you send it, but like anything you type, an unsent draft may be saved in your browser. We keep a usage record of each dictation session, linked to your account (when it happened and how long it ran), with no audio or text.
The browser extension only imports content from supported services or pages when you enable it and choose to connect or capture that content. It does not continuously upload your browsing history, we do not use it to read unrelated browsing activity, and we do not sell extension data.
3. How your content is processed by AI models
- When you send a message, upload a file, generate media, review a document or use another AI feature, the prompt, files and relevant context are sent to the model provider selected by you or by Lippa's Auto routing, solely to produce the requested response. That is processing on your instruction, not training.
- We do not use your content to train Lippa models. For third-party providers, we use API terms, settings and contractual arrangements designed so that prompts, files, context and outputs submitted through Lippa are processed only to provide the requested feature and are not used to train the provider's models. If a feature or integration ever carries different terms, we disclose that where you enable it.
- Your memory layer is stored in Lippa, on EU infrastructure, and is never shared between accounts. Memory items are only sent to a provider when they are selected as relevant context for your own request.
- Some model providers may process data outside the EEA; the model picker identifies each model's provider. Video generation currently uses Google's Veo and may be processed in the United States; this is shown at the point of use and is optional per request. Finished clips are stored in your Lippa library on EU infrastructure.
- Dictation. Google transcribes your speech, acting as our processor. Depending on availability, your audio goes either to Google Cloud Speech-to-Text in the EU, or to Google's Gemini API, which may process it outside the EEA, in any country where Google or its agents maintain facilities. Under Google's terms for this paid service, Google does not use your audio to improve its products. Lippa never writes your audio to disk, a database or our logs.
- We may use automated systems to scan prompts, uploads, generated media, shared links and public pages for security, abuse, child-safety, non-consensual-imagery, fraud, malware, impersonation and other prohibited-content risks. Flagged items may be reviewed by trained staff where necessary. We do not generally monitor private workspace content, but we may review content when it is reported, flagged, required by law, necessary for security, or needed to enforce our terms. Where content is reported or blocked for a safety reason, we may generate and retain a hash (a fingerprint) of that content and use it to detect and prevent re-submission of the same material, and, for the most serious categories, to support reporting to the relevant authorities.
4. Google Workspace data (Google Drive and Google Calendar)
Lippa can connect to your Google account so that you can use your Google Drive files and Google Calendar in Lippa chats. This section explains what we access, what we do with it, and what we never do. It applies only if you choose to connect Google.
What we access
- Google Drive: only the files you pick for Lippa with Google's file picker. We cannot see or search the rest of your Drive.
- Google Calendar: events on your primary calendar. We read the title, time, location, the attendees' display names and the event description.
- We do not request or use the attendees' e-mail address fields.
- An event's description can still contain e-mail addresses, if someone wrote them there.
- Your Google account identifier, and the e-mail address of the connected account when Google provides it, so that we can show you which account is connected.
Lippa only reads. Lippa never creates, changes or deletes anything in your Google account. Google's permission for files you pick technically allows changes to those files; Lippa's systems do not make them, and we do not request any write permission for your Calendar.
How we use it. Only to provide the features you use:
- to answer your questions in a chat about files you picked and about your calendar;
- to attach a Drive file you choose to a message;
- to save a copy of a Drive file to a Space, when you ask us to.
What we never do with Google data
- We do not use it, or anything derived from it, to develop, improve or train artificial intelligence or machine-learning models. That applies to our own models, and we only send it to AI providers whose terms for our accounts exclude such training.
- We do not sell it, use it for advertising, or pass it to data brokers.
- We do not use it to determine creditworthiness, or for lending.
- Our staff do not read it, except:
- when you ask us to and give documented permission for that specific content; or
- when it is necessary for security (for example, investigating abuse) or to comply with the law.
How we share, transfer or disclose Google user data
We share, transfer or disclose Google user data only to the following recipients, and only to provide the features you use:
- The AI model you choose for the chat. Chats that contain Google data can only use models from providers whose terms, for the accounts Lippa uses, exclude training on our data: OpenAI, Anthropic, Google, Mistral and xAI. These providers may keep inputs for a limited time for abuse monitoring, under their own terms. They do not train on them.
- Our web search provider, Tavily, when a web search runs in a chat that contains Google data. A search query can contain text from that data. Our account is set so that Tavily does not save or use our queries. Tavily may pass a query on to another search engine (such as Google) when it needs to in order to answer it.
- The people who can see a Space, if you save a copy of a Drive file to that Space.
- AI tools you have connected to Lippa yourself, such as an assistant using Lippa through MCP, if they read a Space that holds a copy you saved. Those tools process it under their own terms, which you agreed with their provider.
- Our hosting providers, which store and process data on our behalf (see "Where your data lives").
- Authorities, where the law requires it.
We do not share, transfer or disclose Google user data to anyone else. A chat that contains Google data cannot be shared by link or published to others.
Where it is kept, and for how long
| Data | How it is kept |
|---|---|
| The credential that lets Lippa reach your Google account (a refresh token) | Stored encrypted. When you disconnect, Lippa stops using it at once. It is deleted as soon as Google confirms that our access is revoked; if Google cannot be reached, it is kept only for the retries, unusable for anything else |
| Short-lived access tokens | Held in memory only while a request runs, and never stored |
| Google content in a chat, and Drive files attached to a message | Kept as part of that chat. They stay private to you, and are excluded from your Lippa memory, search index and context packs. They are deleted from our live systems when you delete the chat or your account |
| A copy you saved to a Space | Part of that Space: a snapshot that does not follow later changes in Google. It stays until it is deleted from the Space. In a shared Space, deleting your own account does not delete content the Space keeps |
| Backups | Deleted data is removed from our backups on a rolling cycle, typically within 30 to 90 days |
Disconnecting Google stops all further access. It does not by itself delete existing chats or saved copies; delete those separately.
How we protect it
- Data is encrypted in transit and at rest.
- The Google credential is encrypted separately, with keys held in a secrets manager.
- Every request to Google is checked against your current connection, plan and workspace membership.
Disconnecting and deletion
- Disconnect Google at any time in Settings → Apps. We stop using your Google connection immediately, then revoke our access at Google and delete the credential.
- If the same Google account is still connected in another of your Lippa workspaces, we keep that other connection working and do not revoke at Google. In that case, you can remove Lippa's access entirely at myaccount.google.com/permissions.
- Deleting your Lippa account disconnects Google and deletes the data described above, as set out in the table.
Google's policies. Lippa's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements, and to the Google Workspace API User Data and Developer Policy.
Changes. We will tell you, and ask for your consent again before the change applies, if we change:
- what Google data we access;
- what we use it for;
- the kinds of recipients listed above.
The current list of approved AI providers is kept in this section.
5. GitHub data
Lippa can connect to your GitHub account so that you can use your repositories in Lippa chats. This section explains what we access, what we do with it, and what we never do. It applies only if you choose to connect GitHub.
What we access
We access only repositories where the Lippa app is installed and that you can read yourself on GitHub. GitHub checks this on every request. Within those repositories:
- The list of repositories, with their names, branches and tags, so that you can choose from them and the AI model can find them.
- Files, their names and contents:
- files you add to a chat;
- files the AI model reads or searches when your question needs them.
- Commit history: commits, their messages and authors, the files they changed and the changes, and comparisons between branches, tags or commits.
- Pull requests: their titles, descriptions, changes and changed files, their comments and reviews, and the checks and statuses of their commits.
- Logs of failed checks run by GitHub Actions.
- Your GitHub account name, so that we can show you which account is connected.
We do not read issues, discussions or wikis, and from GitHub Actions we read only the logs of failed checks.
Text written by other people. Pull requests, comments, reviews, commit messages and logs can include text written by other people, and their GitHub names. We handle it as this section describes, only to answer your question.
Lippa only reads. Lippa never creates, changes or deletes anything on GitHub. The Lippa app asks GitHub for read access only: to repository contents and metadata, pull requests, checks, Actions and commit statuses.
What we leave out
- Symlinks, submodules, Git LFS files, binary files and very large files are skipped.
- Files likely to hold secrets, such as
.envfiles and private keys, are left out by their names, and so are their changes. - Text that looks like a credential is removed from logs before the AI model sees them, and from the excerpts we keep of what it looked up.
- A secret written inside an ordinary file, comment or description is not always detected.
- When you add files, Lippa shows you what it included and what it skipped.
How we use it. Only to answer your questions in the chat where you added the files or asked the question. The AI model looks things up only when your question needs it, and each look-up shows in the chat as a step, with what was looked up.
What we never do with GitHub data
- We do not use it, or anything derived from it, to develop, improve or train artificial intelligence or machine-learning models. That applies to our own models, and we only send it to AI providers whose terms for our accounts exclude such training.
- It is not added to your Lippa memory, search index or context packs.
- We do not sell it, use it for advertising, or pass it to data brokers.
- Our staff do not read it, except:
- when you ask us to, and give documented permission for that specific content;
- when it is necessary for security (for example, investigating abuse);
- or to comply with the law.
How we share, transfer or disclose GitHub data
We share, transfer or disclose GitHub data only to the following recipients, and only to provide the features you use:
- The AI model you choose for the chat. Chats that contain GitHub data can only use models from providers whose terms, for the accounts Lippa uses, exclude training on our data: OpenAI, Anthropic, Google, Mistral and xAI. These providers may keep inputs for a limited time for abuse monitoring, under their own terms. They do not train on them.
- Our web search provider, Tavily, when a web search runs in a chat that contains GitHub data. A search query can contain text from that data. Our account is set so that Tavily does not save or use our queries. Tavily may pass a query on to another search engine (such as Google) when it needs to in order to answer it.
- Our hosting providers, which store and process data on our behalf (see "Where your data lives").
- Authorities, where the law requires it.
We do not share, transfer or disclose GitHub data to anyone else. A chat that contains GitHub data cannot be shared by link or published to others.
Where it is kept, and for how long
| Data | How it is kept |
|---|---|
| The credentials that let Lippa read on your behalf (an access token and a refresh token) | Stored encrypted. When you disconnect, we ask GitHub to revoke our access and delete them. If GitHub can't be reached, they are deleted anyway, and we tell you how to remove Lippa in your GitHub settings |
| Files you added to a chat | Kept as part of that chat, as a copy at the commit you chose; they don't follow later changes on GitHub. They stay private to you, and are deleted from our live systems when you remove them from the chat, or delete the chat or your account |
| What the AI model looked up in a chat | Kept as part of that chat: the answers, which may quote it; for each look-up, what was looked up and the names it found (such as file paths and pull request titles); and for a file, commit, pull request or log it read, the first 1,200 characters, with text that looks like a credential removed. Private to you, and deleted from our live systems when you delete the chat or your account |
| Backups | Deleted data is removed from our backups on a rolling cycle, typically within 30 to 90 days |
Disconnecting GitHub stops all further access. It does not by itself delete GitHub data already in your chats; delete those chats, or remove the files, separately.
How we protect it
- Data is encrypted in transit and at rest.
- GitHub credentials are encrypted separately, with keys held in a secrets manager.
- Every request to GitHub is made with your own authorisation, and is checked against your current connection and plan. GitHub decides, at that moment, what you can read.
- Repository names, file paths, file contents, pull request text and logs are kept out of our logs and error reports.
Disconnecting and removing the app
- Disconnect GitHub at any time in Settings → Apps. We stop using your GitHub connection immediately, revoke our access at GitHub and delete the credentials.
- Disconnecting does not uninstall the Lippa app from your GitHub account or organisation. To uninstall it, go to your GitHub settings → Applications → Installed GitHub Apps. Organisation owners do this in the organisation's settings.
- Deleting your Lippa account disconnects GitHub and deletes the data described above, as set out in the table.
Changes. We will tell you before we change what GitHub data we access, what we use it for, or the kinds of recipients listed above.
6. Why we process, and on what legal basis
- Running the service you signed up for, including AI processing, storage and sync (contract, GDPR art. 6(1)(b)).
- Security, abuse prevention and safety review (legitimate interest, art. 6(1)(f), and legal obligation where reporting duties apply).
- Billing and accounting (contract and legal obligation).
- Mailing list and marketing emails (consent, art. 6(1)(a), withdrawable any time via the unsubscribe link). Service, security and billing notices are part of providing the service and do not require marketing consent.
- Aggregate, de-identified product analytics inside Lippa (legitimate interest).
- Website analytics and advertising measurement (consent, art. 6(1)(a)), withdrawable at any time via Cookie settings in the footer. We do not build advertising profiles from your use of Lippa itself.
Where we rely on legitimate interests, those interests are operating and securing Lippa, preventing abuse and fraud, enforcing our terms, maintaining reliability, understanding aggregate feature use, and protecting users and third parties. You may object to processing based on legitimate interests; we stop unless we have compelling legitimate grounds or legal claims.
7. Where your data lives, and for how long
- Workspace data is stored and processed on EU infrastructure, with data residency in the EU as the default. Exceptions are described in section 3.
- Account and workspace content is kept while your account is active. Deleted content is removed from live systems promptly and from backups on a rolling cycle, typically within 30 to 90 days.
- Billing, invoice and accounting records are kept for the periods required by Norwegian accounting and tax law.
- Security logs are normally kept up to 12 months, longer only where needed for an investigation, fraud prevention or legal claims.
- Abuse, safety and moderation records are kept as long as needed to enforce our terms, protect users and comply with legal duties, with longer retention for severe abuse, child-safety, fraud, security or legal matters.
- Mailing-list data is kept until you unsubscribe or we delete inactive lists.
- When you close your account, you get a window to export everything; after that, your data is deleted, except the minimum we must keep for the purposes above.
Dictation audio is not stored by Lippa. It may be transcribed outside the EEA (see section 3).
8. Who we share data with
Nobody buys your data. We use processors and service providers to run Lippa:
- AI model providers (currently including OpenAI, Anthropic, Google, Mistral, xAI (Grok), DeepSeek and Zhipu), receiving only the prompts, files, context and outputs needed for your requests, under the no-training arrangements described in section 3. The current roster is always visible in the model picker;
- Google Vertex AI (Veo) for video generation, where you choose it (US, labelled at the point of use);
- Google (Gemini API and Google Cloud Speech-to-Text) to transcribe dictation, which may be processed outside the EEA;
- Tavily for web search, when a web search runs as part of your request;
- Stripe for payment processing and subscription billing;
- EU-based cloud infrastructure and Cloudflare for hosting, storage, security and delivery;
- Google Ireland Ltd for Google Analytics and advertising measurement on this website, only if you accept analytics cookies;
- transactional email, support and mailing-list providers;
- logging and error-monitoring tools used to operate the service, without third-party advertising profiles.
If you connect Google Drive or Google Calendar, we share, transfer or disclose Google user data only as described in section 4, “How we share, transfer or disclose Google user data”.
Where a provider processes personal data outside the EEA, we rely on recognised transfer mechanisms, an adequacy decision or the EU standard contractual clauses, and where required we assess the transfer risk and apply supplementary technical, contractual or organisational measures. Providers based in countries without an adequacy decision are identified as such in the model picker. We disclose data to authorities only where the law requires it or where necessary to protect users or others, and we tell you when we lawfully can.
If you connect GitHub, we share, transfer or disclose GitHub data only as described in "GitHub data".
9. Cookies and local storage
This website sets strictly necessary cookies and, only if you accept, Google Analytics cookies that tell us which pages and campaigns bring people to Lippa. We use Google Consent Mode v2: for visitors in the EEA, the UK and Switzerland, no analytics or advertising storage is used unless you choose Accept, and the same choice governs the measurement of any advertising we run. You can change or withdraw it at any time via Cookie settings in the footer.
Website analytics tells us about pages, referrers and campaigns. It never includes your conversations, files or memory items. The app itself uses functional cookies, local storage and similar technologies for your session and preferences.
10. Your rights
Under the GDPR you can access, correct, delete and export your data, restrict or object to certain processing, object to direct marketing at any time, and withdraw consent at any time. Export and deletion are built into the product; for anything else, write to hello@lippa.ai and we respond within a month. You can also complain to Datatilsynet or your local supervisory authority.
One honest note on deletion: removing a memory item or conversation removes it from future retrieval once deletion is processed, but outputs you already exported or published may remain outside Lippa's control.
11. Children
Lippa accounts require a minimum age of 18. Your date of birth is checked at sign-up but not stored; we keep only a record that the requirement was met. We do not knowingly allow anyone under 18 to use Lippa. If you believe a child has created an account or used a restricted feature in breach of these rules, contact us and we will take appropriate action.
12. Changes
If this notice changes in a way that matters, we tell you in the product or by email before it takes effect. The latest version always lives at this address.
Questions about privacy: hello@lippa.ai