Privacy policy
The Shenas privacy policy. It explains the small amount of data Shenas servers hold about you, why we hold it, what we do with it, and how you can control it. The most important thing is in §2: your personal data does not live on Shenas servers.
This is our privacy policy. It explains the small amount of data Shenas servers hold about you, why we hold it, what we do with it, and how you can control it.
It is written for you, the person whose data this is. It is not written for lawyers. Where European law requires us to use a specific phrase, we use it once, plainly, and we explain what it means.
The most important thing about Shenas's privacy posture is at the top, in §2, because it is unusual and we want you to read it: your personal data does not live on Shenas servers. shenas — the software that holds your personal data — runs on your own hardware. We provide a small set of services around it (an account, a community chat, and a way for your own devices to find each other), and that is all.
If something here is unclear, write to us. Our address is in§1.
1. Who we are
The company responsible for this service is shenas ai, Inc., a Delaware corporation. For postal correspondence about this policy, use our registered agent:
- Email: privacy@shenas.ai
- Postal: shenas ai, Inc., c/o Legalinc Corporate Services Inc. (Registered Agent), 131 Continental Drive, Suite 305, Newark, DE 19713, United States, Attn: Privacy.
We use our Delaware registered-agent address as the public mailing address for privacy correspondence so that mail to "shenas ai, Inc., Attn: Privacy" reaches us through a stable corporate channel rather than a residential address.
We have not yet engaged an Article 27 representative in the Union. If you are an EEA or UK resident and need to contact a representative directly, please contactprivacy@shenas.ai in the meantime, and complain to your local supervisory authority via §7 if needed.
In European law terms, we are the data controller for the data described in this policy. That means we decide what data is collected and what it is used for, and we are accountable for those decisions.
2. The most important thing — what we do not have
Shenas's architecture is deliberately narrow. The personal data you keep in shenas — your calendar, your email, your activity records, your photos, your notes, anything your shenas node has sourced from a third-party service like Google, Strava, Withings, Garmin, Spotify or any other — lives only on the hardware you control. Not on Shenas servers. Not on any server we operate. We do not see it. We do not store it. We do not have access to it.
shenas is a personal-analytics tool. When you connect a third-party service (for example Strava) to shenas,your node sources a copy of your data from that service into your own storage and uses it locally for analytics on your behalf. The data is not flowing between services — it is pulled from each service you choose into one place that you control, where you can look at it. The access token is held by your node. The data is held by your node. Shenas is not in the path. We do not need a "we are sorry we got hacked" letter ready for your Strava data, because we do not have your Strava data.
What Shenas servers do hold is described in§3 — and it is much less than you might expect.
3. What this policy covers, and the small set of data we do hold
This policy applies to the Shenas centralised services:
- The Shenas account at
auth.shenas.net, which is your single sign-on to the two services below. - The community chat at
matrix.shenas.net, where Shenas users and developers talk to each other. - The device registry, which records the shenas devices you register against your account. It holds no network addresses and no message contents; see §3.3.
- The peer directory, which is not switched on yet. When it is, it will hold the network addresses your devices can currently be reached on while they sync with each other, so they can find each other; see §3.3.
- The marketing site at
shenas.ai. - Customer support correspondence you send us.
This policy does not apply to:
- Your shenas data. shenas runs on your hardware. It sources data from the services you connect, stores it on your hardware, and uses it for analytics on your behalf. None of that is sent to us. The data on your node is governed by your own choices and by any disclosures the shenas software shows you at the time you grant a third-party connection.
- The open-source Shenas project at
shenas.org. The OSS project does not, by itself, collect personal data from end users. - Third-party services your shenas node sources data from (Strava, Google, Spotify, Withings, Garmin, and others). Those services have their own privacy policies and process your data under those policies, not ours.
When you export some of your shenas data to share it with another person, the file does not pass through us either; see §3.7.
3.1 Your Shenas account
We hold:
- Your email address, used to identify you and to send security and account messages.
- Your chosen display name for community chat.
- A link to your profile picture, if your Shenas account has one. We keep the web address of the picture, not the picture itself, and copy it from your account each time you sign in.
- A salted hash of your password (never the password itself). If you sign in via a third-party identity provider, we hold the linkage but not their credentials.
- Your account identifier: the permanent ID our sign-in service at
auth.shenas.netgives your account. The Shenas application uses it to recognise you each time you sign in, and it stays the same if you change your email address. It is also written to the working logs of the server that handles your sign-in. - Your plan and access: which hosted features your account may use (for example a paid plan, or connecting your devices so they sync), how that access was granted and until when, and whether yours is a staff account. We keep what is in force now, not a history of it.
- Multi-factor authentication enrolments if you enable them (the registered factor, not the secrets you keep on your device).
- Sign-in event records — a record of each successful sign-in to your account (when it happened, and whether it signed you in, linked an existing account, or created a new one), kept for security and to help you spot abuse of your account.
Your password hash and multi-factor enrolments are stored at auth.shenas.net, which also holds your email address, display name and account identifier. The Shenas application keeps its own copy of your email address, display name, profile-picture link and account identifier, together with your plan and access and your sign-in event records, in its own datastore.
3.2 Community chat (matrix.shenas.net)
The community chat is a Matrix server. When you participate, we hold:
- The messages you send in public rooms and in private rooms you participate in.
- Room and channel membership — which rooms you are in.
- Presence and typing indicators while you are connected.
- Display name and avatar you set for chat.
Matrix is a federated protocol. If you join a room that includes users on other Matrix servers, your messages reach those servers as part of the protocol. We cannot retract messages from other servers; you can ask them directly, and we can help with the request.
3.3 The device registry
You register a device by enrolling it for a hosted feature on your account — turning on the LLM analysis or literature review features on an install, for example — not by signing in. Signing in proves which account is asking; enrolling is the separate step that adds the device to this registry. We hold one record per enrolled device:
- A device identity record — the public key you generated on the device, the device name you chose (which defaults to a randomly generated, non-identifying string like "calm-ridge-47"), the device type (laptop, desktop, mobile, server), which hosted features it is enrolled for (today, only the LLM analysis and literature review features; the relay and the mailbox are built but not offered yet), and the time we last saw the device.
That is the whole record. The device registry holdsno network addresses; those are in the peer directory below, kept apart from it. We do nothold any record of which of your devices has synced with which, or any message your devices exchange. Your devices sync directly, over connections that do not pass through us.
The peer directory. The peer directory is not switched on yet. When it is, each of your devices that syncs with the others will tell us the network addresses (IP address and port) it can currently be reached on, so your other devices can find it without you copying an address by hand. We will hold one record per device: those addresses, the device's network identifier, and which account it belongs to. Each record will be replaced whole every time the device checks in, and will expire one hour after the device last checked in. A device on which you have not turned on pairing will publish nothing. Only your own devices, signed in to your account, will be able to look up your records; no other Shenas user will. We will keep no history of where a device has been, no record of which device looked up which, and no record of when your devices synced. We will not switch it on while its requests pass through Cloudflare (see§8), and we will update this policy before we do.
Until then, devices with pairing on already try to reach it. If you have turned on pairing in the Shenas app (version 1.6.0 or later), the app contacts the peer directory every ten minutes, sending the network addresses the device can currently be reached on and asking for those of your other devices. Our servers answer that the directory is off. They note that a request arrived, without the address it came from or the account it belongs to, and keep nothing it sent. But every request passes through Cloudflare first, and Cloudflare's record of it keeps the IP address it came from (see §8). For most devices that is the public address of the network the device is on. So for as long as pairing is on, that record shows, about every ten minutes, which network your device was using. We are changing the app so that it stops making these requests while the directory is off.
What a hosted feature sees when an enrolled device uses it. For LLM analysis, an enrolled device trades proof that it holds its own key for a short-lived token scoped to that one feature. The token says which feature it is for; it does not say which device or which account asked for it, and we do not record which token went to which device. Literature review does not use that token yet: your device signs in to it with your Shenas account, so we do learn which account asked for a literature review. Like any internet service, it also sees the network address your request comes from. If we are ever ordered to start logging that link, we will say so publicly as far as the law allows, and we will move the LLM analysis and literature review features, as soon as we lawfully can, to a token design under which we never learn which token went to which device, rather than a promise we ask you to trust.
What a hosted feature receives. Each request carries only what that feature needs to answer it:
- Literature review receives the kinds of data you keep (such as "sleep" or "spending"), which research findings you already have, and two settings for how many papers to look at. It never receives the data itself. To answer, our server searches two public research databases, OpenAlex and Semantic Scholar, using fixed search terms for each pair of those kinds (for example, the terms for "sleep" together with the terms for "spending"). Those searches come from our server, not your device, and carry nothing that identifies you. We then send the public titles and abstracts we find to our AI provider, Anthropic, PBC, in the United States. Anthropic receives only those public texts, nothing about you.
- LLM analysis receives the request your device builds, which can include some of your data. The request goes to an inference service we run on a server in Germany, which passes it to Anthropic, PBC in the United States to get the answer. That service keeps logs of the requests it handles. We have not yet confirmed what those logs contain or how long they are kept; we will say so here when we have.
What we have built and not switched on. Arelay that would carry traffic between two of your devices that cannot reach each other directly is not deployed and holds nothing. We will update this policy before it is turned on.
Who can see your device records. You, via your own Shenas account, and Shenas operators investigating an abuse or operational incident — including through an internal support view that lists registered devices across accounts. That view shows device records only: no addresses, and no record of which devices sync with each other. Your device records are not visible to other Shenas users.
3.4 Marketing site (shenas.ai)
The marketing site is static. Our own servers keep no log of requests to it. We do not use third-party analytics or advertising trackers. Cookies are covered in §13.
Every request to shenas.ai passes through Cloudflare, Inc. on its way to our servers. That covers the pages of this site and also the requests the Shenas app makes to your Shenas account. Cloudflare sees your IP address and the content of each request in transit. Cloudflare also keeps a record of each request, including your full IP address, for at least 31 days (see§8 and §5).
When the peer directory is switched on, it will also hold the actual network addresses of your devices that sync with each other (see §3.3).
3.5 Customer support correspondence
Anything you send us by email, in-product message, or other channel — including attachments — we hold for as long as we need to resolve your issue (see §5).
3.6 What Shenas servers do NOT hold
To be explicit, because it is the point of the architecture:
- No data your shenas node has sourced from third-party services. Not your Google Calendar, not your Gmail, not your Strava activities, not your Withings weight, not your Garmin heart rate, not your Spotify history, none of it. Your node sources it directly from each service and analyses it on your hardware; we never see it.
- No copy of anything your devices send each other, because none of it reaches us. We operate no relay: your devices sync directly, over connections that do not pass through our servers. We hold neither the contents nor a record that an exchange happened.
- No backups of your shenas data.
If we ever change that — for example, if we offer an optional cloud backup service in the future — we will update this policy, ask you to opt in before the change takes effect for you, and describe the new processing in detail.
3.7 Sharing your data with another person
shenas lets you share some of your data with another person: a partner, a friend, a trainer. This section describes what a share is and what happens to the data. It covers sharing by exporting a file. When you export a share, Shenas, the company, is not part of it: we do not see what you share, who you share it with, or when, and we keep no record of it.
What a share is.
- You choose the person, and you choose what they get. A share is something you set up deliberately, on its own screen. You pick the kinds of data (each one has its own switch), the time range, and whether the other person sees every record or only summaries. Starting points such as "Partner" or "Trainer" fill these choices in for you, and you can change any of them. The Partner starting point leaves out where you have been; you can switch that on yourself.
- An export carries data about you, and shenas leaves out what it can tell is about someone else. Records about another person in your shenas are left out, and so is a place linked to another person. Free-text notes, titles and descriptions are never included, because they often name other people and shenas cannot tell when they do. Your passwords and the logins you use to connect other services are never included.
- Before anything leaves your device, shenas shows you what will be sent and to whom, and asks you to confirm with your device's own unlock (your passcode, fingerprint or face) where your device has one.
- shenas keeps a record on your device of what you agreed to, so you can see what was shared, when, and on what terms. You find your shares in the share list in shenas. The person you share with sees those terms in the notice that comes with the file (see below).
Exporting to a file.
- An export is a file you hand over yourself. shenas writes it to a place on your own device that you choose. It is never uploaded to, stored on, or passed through any Shenas server. How the file reaches the other person (email, a USB stick, a messaging app) is up to you, and shenas does not know whether it arrived. Your share list says "exported on" a date, not "delivered".
- Anyone who has the file can read it. shenas cannot check who ends up with it. The export screen tells you whether the file is encrypted; today it is not.
- An export cannot be recalled. Once the file has left your device, neither you nor we can take it back or delete it from someone else's computer. You can still tell the person you gave it to that you no longer agree to them using it, and the notice that comes with the file asks them to stop when you do. Whether the law makes them stop depends on who they are: a friend or partner who keeps it for private use is not bound by data-protection law, while someone who holds it for their work, such as a trainer, has legal duties of their own about it (see "What the person receiving it is told" below).
Passing your data on.
- A file you export carries no restriction. Whoever has it can pass it on.
- Where shenas marks data as not to be passed on, only the shenas software we publish obeys that mark. It cannot stop someone from taking a screenshot, copying the data, or using another program or a modified version of shenas. We do not call shared data "secure" once it has left your device, because after that it is in the other person's hands.
What the person receiving it is told.
- Every export comes with a short notice in plain language, beside the data. It says what the data is and whose it is, that it cannot be recalled, that any limit on passing it on is enforced only by shenas software, and that it asks them to stop using the data if you tell them you no longer agree.
- The notice also says: if you receive this data for your work, you are responsible for it under data-protection law and the rules of your profession. A trainer or other professional who receives your data this way answers to you for it, and you can use your data-protection rights (see §7) against them directly. We do not hold your data, so we cannot answer for it.
Because an export never reaches us, there is nothing about it for us to hand over if we receive a legal order (see§10), and nothing for a breach of our systems to expose.
3.8 Fediverse server (social.shenas.net)
If you create an account on our fediverse server, we hold your handle, the posts and media you publish, and who you follow and who follows you. Posts you publish publicly are sent to other fediverse servers, which we do not control. We have not yet set how long we keep these; we will put the period here when we have.
4. Why we are allowed to process the data we do hold
European law requires us to identify a lawful basis for each thing we do with your data. Here is the map for the small set of data in §3.
| What we do | Why we are allowed to do it |
|---|---|
| Run your Shenas account; let you sign in to chat and to register devices | We need to do this to deliver the service you signed up for (GDPR Article 6(1)(b), contract). |
| Operate the community chat | We need to do this to deliver the service you joined (Article 6(1)(b), contract). |
| Operate the device registry for your own devices | We need to do this to deliver the service you signed up for (Article 6(1)(b), contract). |
| Operate the peer directory, once it is switched on, so your devices that sync with each other can find each other | We need to do this to deliver the device sync you turned on (Article 6(1)(b), contract). |
| Reply to your support emails | Article 6(1)(b), contract. |
| Keep sign-in event records and audit records | We have a legitimate interest in protecting the service and our users against abuse and attack (Article 6(1)(f), legitimate interests). Our balancing test is on file and available on request. |
| Comply with a valid legal order | We are required to (Article 6(1)(c), legal obligation). See §10. |
| Send you transactional messages (e.g. password reset, security alert) | Article 6(1)(b), contract. |
| Send you marketing about new features (if you opted in) | You asked to receive these (Article 6(1)(a), consent). You can stop them with one click. |
We do not rely on legitimate-interest grounds for anything other than security and abuse prevention. We do not rely on consent for anything we could do under contract.
There is no entry in this map for "process the data you have on your shenas node" or "process data you import from third-party services" because Shenas servers do not do those things. See §2 and the list in§3.6.
5. How long we keep the data we do hold
We keep your data only as long as we have a reason to.
- Account data — for as long as your account exists.
- Community chat messages in public rooms — kept for as long as the room exists, so that scrollback is available to other members. You can request deletion of your own messages at any time; we will action it in our local store, and we will pass the redaction request to other federated servers (we cannot guarantee they will honour it, but we will try, and we will tell you what happened).
- Community chat messages in private rooms — kept for as long as the room exists. Same deletion path as above.
- Device identity records (public key, device name, type, last-seen) — kept while the device is registered; deleted when you deregister the device.
- Device addresses, once the peer directory is switched on — when your devices sync with each other, each one will tell us where it can currently be reached on the network, so the others can find it. We will keep only the latest answer. We will stop using it one hour after a device last checks in and delete it shortly after (within about 15 minutes). We will keep no history of where a device has been, no record of which device looked up which, and no record of when they synced. A daily backup of our database may contain whatever was current at the moment it ran; those backups roll off on the schedule below. Until it is switched on we hold no device addresses in it (see §3.3 for what Cloudflare keeps meanwhile).
- LLM analysis request logs — kept by the inference service that handles LLM analysis for us (see §3.3). We have not yet confirmed how long; we will put the period here when we have.
- Fediverse server accounts — your handle, the posts and media you publish, and who you follow and who follows you (see §3.8). We have not yet set how long we keep these; we will put the period here when we have.
- Sign-in event records and security audit logs — 13 months, then deleted, unless we are required to keep them longer by law.
- Support correspondence — 24 months from the last reply, then deleted.
- Cloudflare's record of requests to
shenas.ai— covering both the site and the app's requests to your account, including the requests described in §3.3: kept by Cloudflare for at least 31 days. Cloudflare does not state a maximum, and we cannot delete entries from it (see §8). - Billing and tax records (if you are on a paid plan) — for the period required by United States federal and Delaware state tax law (typically 7 years).
- Backups of the data above — backups roll off on their own schedule (no more than 35 days). When you delete data, it is gone from active systems as soon as we carry out your request (within the time limits in section 7) and from backups within 35 days of that.
- Marketing suppression token — if you opt out of our marketing email, we keep a one-way hashed token (no name, no email address, no message content) so that we never email you again. We keep it even when you ask us to erase your other data, and we use it for nothing else.
If a legal order requires us to preserve specific data for longer, we do so for the scope of the order and no longer.
6. Your controls
You are in charge of your data.
- In your Shenas identity account: edit your profile, change your password, and enrol or remove MFA. These are handled by our self-hosted identity provider at auth.shenas.net, not by a settings page on this site.
- In the product: deregister your devices.
- In chat: you can redact your own messages and leave rooms from any Matrix client. We will help with cross-server redaction requests where possible.
- Export. Get a copy of all the data we hold about you in a structured, machine-readable form.
To disconnect a third-party service like Strava from yourshenas node, use the shenas software on your node — the Shenas account does not control those connections, because we do not hold them. If a third-party-service control is broken or unclear in shenas, the project's issue tracker at shenas.org is the place to flag it.
If a control in the Shenas account or chat is broken or unclear, emailprivacy@shenas.ai and we will fix it and reply.
7. Your rights under European law
If you are in the EEA, the UK, or another jurisdiction with similar law, you have the following rights, no matter what the product UI offers:
- Access — ask us what data we hold about you, and get a copy (GDPR Article 15).
- Rectification — ask us to correct data that is wrong (Article 16).
- Erasure — ask us to delete your data ("right to be forgotten") (Article 17).
- Restriction — ask us to stop using your data while we resolve a question about it (Article 18).
- Portability — get your data in a machine-readable format, or have us send it directly to another service (Article 20).
- Object — object to processing we do under legitimate interests (Article 21).
- Withdraw consent — for anything we do under consent, change your mind at any time (Article 7(3)). Withdrawal does not affect what we did before you withdrew, but stops anything ongoing.
- Not be subject to automated decisions — we do not make automated decisions about you that produce legal or similarly significant effects (Article 22). If we ever do, we will tell you and you will have the right to a human review.
These rights apply to the data Shenas holds, as listed in§3. Because Shenas does not hold the data on your shenas node, an access or erasure request to us will not retrieve or delete it. To exercise those rights against the data on your shenas node, use shenas directly.
To exercise any of these rights with respect to Shenas-held data, emailprivacy@shenas.ai. We will reply within 30 days. If the request is complex, we may extend by up to 60 more days and tell you why.
You can also complain to a supervisory authority. Because shenas ai, Inc. is established in the United States (Delaware) and does not currently have an EU establishment, there isno single EU lead supervisory authority for our processing under the GDPR's one-stop-shop rule. You can complain to the supervisory authority in your own EU country; the full list is atedpb.europa.eu/about-edpb/about-edpb/members_en. In the United Kingdom you can complain to the Information Commissioner's Office atico.org.uk.
Residents of California and other US states with similar laws
If you are a California resident, you have similar rights under the California Consumer Privacy Act / California Privacy Rights Act: to know, to access, to delete, to correct, to opt out of sale or share (we do not sell or share your personal information), and not to be retaliated against for exercising those rights. To exercise them, emailprivacy@shenas.ai.
We do not "sell" personal information under California law. We do not "share" it for cross-context behavioural advertising.
Residents of other US states with comprehensive privacy laws (currently including Virginia, Colorado, Connecticut, Utah, Texas, Oregon, and a growing list) have similar rights. The same email address is the place to send the request.
8. Where the data we hold is stored
Most of our production services run on Google Cloud Platform (Google Kubernetes Engine and Cloud SQL) in the United States, and their backups are kept there too. One production service, the one that handles LLM analysis (see§3.3), runs on a server in Berlin, Germany. So does our development and pre-productionenvironment. We intend that environment to hold no real end-user data, but we have not yet checked that it holds none; we will say so here when we have. We are evaluating a Swiss managed-Kubernetes provider for a future move of the production tier.
Each night the Berlin server is backed up to storage run by Hetzner Online GmbH. Where that storage is located is currently unknown to us; we will name the country here once we have confirmed it.
Who runs these servers. We set up and administer our Google Cloud project and the Berlin server ourselves, using Stakhanovites, a platform that shenas ai, Inc. develops. Stakhanovites is not a separate company. Administering these servers means we can technically reach the data on them. Google, Hetzner, Anthropic and Cloudflare process data for us directly, as service providers to shenas ai, Inc.
Cloudflare. Cloudflare, Inc., a United States company, sits in front of shenas.ai. Every request to the site, and every request the Shenas app makes to your Shenas account, reaches Cloudflare first, and Cloudflare sees its content before passing it to our servers. Its role is to deliver the site and protect it from attacks. Cloudflare keeps a record of each request: when it was made, which page or address onshenas.ai it asked for, how we answered, and the IP address it came from. Cloudflare keeps this record for at least 31 days. It does not say how much longer, and we cannot switch the record off or delete entries from it.
When data has to move from the EEA, the UK or Switzerland to the United States, we use theStandard Contractual Clauses approved by the European Commission (Decision 2021/914). For UK data we add the UK International Data Transfer Addendum. For Swiss data we add the Swiss addendum, which makes the adaptations the Swiss Federal Data Protection and Information Commissioner requires. These clauses also cover the requests Cloudflare handles for us. Its data processing agreement includes them. We have assessed the transfer of our edge access logs to the United States. The assessments for the rest of our Google Cloud processing and for Cloudflare are in progress. We will update this section when they are done. You can ask us for a redacted copy of any finished assessment.
Companies that handle Shenas-held data on our behalf must be bound by a written agreement that holds them to the data-handling commitments in this policy. We will publish our processor list at shenas.ai/legal/subprocessorswhen that page is live, and will update this policy to add the link before any processor change takes effect. We update that list before adding a new processor, and you can object before the change takes effect.
The data on your shenas node is stored on your own hardware in a location you choose. That part is not on us.
9. What we never do with your data
These commitments cover both the small amount of data Shenas servers hold and, where we have a way to enforce it, the architecture choices that keep your shenas data off our servers in the first place.
- We do not train artificial-intelligence or machine-learning models on your data. Not on chat messages, not on discovery metadata, not on anything else we hold. We do not train models on shenas data either, because we do not have it.
- We do not sell your personal data. No exceptions.
- We do not share your personal data for advertising. Not cross-context behavioural advertising, not contextual ad targeting, none of it.
- We do not analyse the data we hold for marketing or customer-insight reports we sell to third parties.
- We do not profile you for advertising purposes anywhere.
- We do not silently expand what we collect. If we decide we need to add a new processing activity that affects you, we will update this policy, give you advance notice, and where European law requires it, ask for your consent before the change takes effect.
If you find a place where our practice does not match this list, write to us. That is a bug, not a policy.
10. Government and law-enforcement requests
When law enforcement, a court, or a government agency asks us for data we hold, our default is:
- We require valid legal process. A court order, a search warrant, or the equivalent under the law of the requesting jurisdiction. We do not honour informal requests.
- We respond with the minimum data required by the order. No more.
- We notify you that the request happened, unless we are legally prohibited from doing so. When the prohibition lifts, we tell you.
- We do not proactively share data with governments. We do not run programmes that send data to law enforcement before they ask.
- We push back on requests that are over-broad, vague, or unlawful. Where we have grounds to object, we object.
- We can only produce what we actually hold. Data that lives on your shenas node never reaches our servers, so for that we have nothing to give and will say so. Your devices sync directly with each other, over connections that do not pass through us, so we hold neither what they exchange nor any record that they did. What can be compelled from us is the small set in §3: your account, your device identity records, community chat in rooms we host, support correspondence, and Cloudflare's record of requests to
shenas.ai, which we can read (see §8). Cloudflare is a United States company and can also be asked for that record directly. Once the peer directory is switched on, the network addresses your paired devices last published will be on this list too, for as long as §5 says we keep them.
We are working on a public transparency report that will describe (in aggregate, without identifying you) the requests we receive and how we respond. When it is live, you will find it at shenas.ai/legal/transparency; we will update this policy to add the link when that page is published.
11. Security
- All connections to Shenas services use TLS 1.2 or better. Data Shenas servers hold is encrypted at rest (AES-256 or equivalent).
- Authentication is handled by a dedicated identity service (Kanidm at
auth.shenas.net) that supports strong second factors. - Access to production systems is limited to a small number of named operators, governed by least privilege, and audited.
- Your devices sync directly with each other. We operate no relay that holds or forwards their traffic, so there is no buffered payload for an operator to read or for a breach to expose (see §3.3).
- We test our defences and patch promptly. We do not claim that any system is unbreakable; nothing on the internet is.
If something does happen — a breach that puts your Shenas-held data at risk — we will tell you. European law gives us 72 hours from awareness to notify the supervisory authority of a personal-data breach; we will also notify you directly without undue delay when the breach is likely to result in a high risk to your rights and freedoms. We will say what happened, what data was affected, what we have done, and what you should do.
A breach of your shenas node is something we cannot detect for you, because we are not in the path. The shenas software is built to help you secure that node; please do not skip the parts of the documentation that explain how.
To report a security vulnerability in Shenas services, seegithub.com/shenas-org/shenas/security/advisories/newor emailsecurity@shenas.ai.
12. Children
Shenas is not intended for users under 18. We do not knowingly create accounts for anyone under 18. If we learn that we have collected personal data from someone under 18, we will delete it.
If you are a parent or guardian and believe your child has signed up for Shenas, contactprivacy@shenas.ai and we will resolve it.
13. Cookies and similar technologies
The marketing site at shenas.ai uses onlystrictly necessary cookies — the cookies the site needs to load, route requests, and remember whether you accepted this notice. We do not use advertising cookies, tracking pixels, or third-party analytics on the marketing site.
The community chat and account services store a session token after you sign in. That is what lets you stay signed in. No third-party trackers run inside our services.
If the marketing site starts using non-essential cookies in the future, we will publish a separate cookies notice and ask you to accept them before they run.
14. Changes to this policy
When we change this policy in a way that materially affects you, we will:
- Send an email to the address on your account.
- Show a notice the next time you sign in.
- Give you at least 30 days' notice before the change takes effect, unless the change is required immediately by law.
When the change is purely editorial (typos, clarifications), we will update the date at the top and add an entry to the revision history below.
Revision history
| Version | Date | Summary |
|---|---|---|
| v11 | 2026-10-30 | Correction. Versions 5 to 10 said the peer directory was switched on from 2026-10-11. It was never switched on, and it never stored a record. But devices on which pairing is turned on (Shenas app 1.6.0 or later) have been contacting it every ten minutes since then, sending their network addresses. Our servers answered that it was off and kept nothing those requests sent; Cloudflare's record of the requests kept the IP address each came from. So for those devices the earlier promise that we keep no history of where a device has been did not hold. §3, §3.3, §4, §5 and §10 now describe the directory as not yet switched on, and say what a device with pairing on sends meanwhile. §3.4, §5 and §8 say what record Cloudflare keeps of every request to shenas.ai, and for how long. §3.4 and §5 no longer describe a 30-day web-server log for this site: we do not run one, and Cloudflare's record is the only record of each request. Nothing changes in what we do; this version corrects what we said. |
| v10 | 2026-10-22 | §3.4 and §8 name Cloudflare, Inc., a United States company. Every request to shenas.ai, including the requests the Shenas app makes to your account, passes through Cloudflare. That was already the case; earlier versions did not say so. §8 adds that the Standard Contractual Clauses cover Cloudflare, and that its transfer assessment is in progress. Nothing changes in what Shenas servers hold. |
| v9 | 2026-10-22 | §3.1 adds three things we already held but did not list: a link to your profile picture, the permanent account identifier our sign-in service gives you, and your plan and access. It also says that the Shenas application keeps its own copy of your account details. We hold more than the previous version said. §8 says that one production service and our development environment run on a server in Berlin, Germany; that the server's nightly backups go to Hetzner storage whose location we have not yet confirmed; that we administer our servers ourselves, using Stakhanovites, a platform shenas ai, Inc. develops, and that Google, Hetzner and Anthropic process data for us directly; and that we have not yet checked that the development environment holds no real end-user data. The previous version said it holds none. New §3.8 describes our fediverse server, and §5 says we have not yet set how long it keeps your data. |
| v8 | 2026-10-19 | §8: names the Swiss addendum for Swiss data, and corrects what we said about transfer impact assessments. The previous version said we had one for each receiving country; we have finished one, for our edge access logs, and the one for the rest of our Google Cloud processing is in progress. Nothing changes in what Shenas servers hold or where. |
| v7 | 2026-10-18 | New §3.7: what a share is when you share your shenas data with another person by exporting a file, what the person receiving it is told, and that an export never passes through Shenas servers. Nothing changes in what Shenas servers hold. |
| v6 | 2026-10-12 | §3.3: describes the service tokens enrolled devices use, and that we do not record which token went to which device. |
| v5 | 2026-10-11 | The peer directory is switched on. §3.3 describes it in the present tense and what it holds; §3 lists it; §3.4 says it holds actual addresses, not a region; §4 adds its lawful basis; §5 adds how long device addresses are kept, including in backups; §10 adds those addresses to what can be compelled. The relay is still not deployed. We hold more than the previous version said: the current network addresses of your devices that sync with each other. |
| v4 | 2026-10-10 | §3.3: says that enrolling a device for a hosted feature, not signing in, adds it to the registry, and lists the features each device is enrolled for. |
| v3 | 2026-09-10 | Corrected §3.3, §3.6, §5, §10 and §11: the relay, the stored relay payloads, the reachability endpoints, the sync cursors and the connection-attempt logs they described no longer exist, and the services that held them were removed. We hold less than the previous version said we did. |
| v2 | 2026-07-26 | Relay message retention reduced from 30 to 7 days. |
| v1 | 2026-05-17 | Initial draft revision (subject to change). |
15. How to reach us
For anything in this policy, including exercising your rights:
- Email: privacy@shenas.ai
- Post: shenas ai, Inc., c/o Legalinc Corporate Services Inc. (Registered Agent), 131 Continental Drive, Suite 305, Newark, DE 19713, United States, Attn: Privacy.
- Security disclosures: security@shenas.ai or GitHub private advisory.
- Point of contact for the community chat (EU Digital Services Act, Articles 11 and 12): Authorities in the EU and the UK, and anyone using the services, can reach us about the chat at info@shenas.ai, in English.
- In the EEA / UK, our Article 27 representative: We have not yet engaged an Article 27 representative in the Union; in the meantime, please contact privacy@shenas.ai or your local supervisory authority (see §7).
We aim to reply within five working days, and always within 30 days for formal requests under the rights listed in§7.