My Sign asks for the moment you were born and then holds what you tell it about your life. That is a lot to trust something with. This page says exactly what happens to it, in plain language, and every sentence here is meant to describe something the app actually does. If one stops being true, the page is wrong and we will fix it.
Who we are. My Sign is operated by [operator name and address — to be completed before public launch]. Questions about this page go to support@mysignapp.com.
Your email address — to sign you in. We send a six-digit code; there is no password.
Your name, birth date, birth time and birth city — to compute your chart. If you skip the time we say so on the reading rather than guess.
Your timezone and country — your timezone comes from your browser, so "today" and the morning notification follow where you are. Your country decides which crisis line you are shown.
Daily check-ins — a mood number from 1 to 10 and, if you add one, a line of text.
Your conversations — everything you type and everything the app replies, including messages sent at a difficult moment.
What the app remembers — at the end of each day, a short summary of that day, a few of your exact words worth keeping, the names and topics that came up, and the mood. Once there is enough history, it also keeps patterns it has checked against your days. Your questions are tagged by topic so it can notice what you keep coming back to.
Dated life events — only if you add them to narrow down an unknown birth time.
Other people — see Other people's details, below.
A notification subscription — only if you turn on the morning notification. It is a device address and encryption keys issued by your browser or phone.
Replies you report — if you flag a reply as wrong or hurtful, we keep which reply and the reason you give.
Usage bookkeeping — how many messages you have sent today (for the free-plan and safety limits), whether a reply referred back to something you said and how quickly you answered it, and which daily lines you have already seen so they are not repeated. Sign-ups and returns are also counted as daily totals that contain no names or IDs.
We do not collect your location, contacts, photos, or advertising identifiers. The app sets one cookie, which keeps you signed in. It does not use analytics or tracking cookies.
What we do with it
Your chart and your conversations are the product. The app reads today's sky against your birth chart, remembers what you have told it, and uses both to answer you. Memory is the point: the app keeps what you say so that it can refer back to it later. That is stated on the consent screen because it is unusual and you should choose it deliberately.
We use your data to run the app for you and to keep it safe. We do not use it for advertising, we do not sell it, and we do not use it to build a profile of you for anyone else.
The AI we use, and what it sees
My Sign's replies, readings and memory are written by a third-party AI model: Claude, made by Anthropic, PBC. To do that, the app sends some of your personal data to Anthropic over its commercial API. Here is exactly what goes, and when:
When you send a message: the message itself, roughly your last ten messages and replies, today's check-in (mood and text), your chart facts and today's sky, short summaries and a few of your own quoted words from recent and related days, the names and relationships of people you have mentioned or saved, and their compatibility scores. If the AI needs more, it can look up your chart, the sky on another date, a saved person's compatibility report, or your past days.
Before that, a safety check: the text of your message, on its own, is sent to the same model to judge whether it needs a careful or crisis response. If a message clearly signals crisis, the app answers with a crisis line straight away. In that case the message is not sent to the AI for a reply.
Once a day: the previous day's whole conversation and check-in are sent, so the AI can write that day's memory.
Once a week, once you have ten or more days: your saved daily memories are sent so the AI can suggest patterns. Our own code then checks each one against your days before you see it.
Once per reading: your first name and chart facts for your birth-chart reading, and a saved person's name and computed compatibility for their written reading.
Morning notification: by default the morning line is chosen from lines we wrote, with no AI involved. If you switch to the sky-based notification, the same kind of context as a chat message is sent each morning to write it.
Never sent to Anthropic: your email address, your account ID, your notification subscription, or your raw birth date, time and place. Only the computed chart is sent.
What Anthropic does with it. Anthropic processes this data to produce the response and under its commercial terms. By default it does not use API data to train its models. It deletes API inputs and outputs from its systems within 30 days. The exception is a request its automated systems flag as violating its usage policy: those inputs and outputs can be kept for up to 2 years. See Anthropic's privacy center: how long data is stored and whether data is used for training.
Before the app stores or sends anything about your day, it asks you to tick a box that names Anthropic and agrees to this. If you do not want your data sent to Anthropic, do not use the chat, readings or memory. Those are the features that need it, and without them the app cannot work. You can delete everything at any time (see Your controls).
Who else sees it
Amazon Web Services stores and runs everything, in the US (Ohio, us-east-2): the database, the servers, backups, logs, and sign-in emails, which go through Amazon SES. Data is encrypted in transit and at rest.
Anthropic, as described above.
When a message suggests someone may be in danger, a person. The app replies with a crisis line for your country. At the same moment it alerts the on-call member of our team through PagerDuty , an incident-alerting service. The alert contains your account ID and the first 120 characters of that message, so a human can judge quickly how serious it is. That person may then read your conversation. We do this because missing someone in danger would be worse. It is not an emergency service, and nobody will contact emergency services on your behalf.
During the pilot, a member of the team. Our team does not read conversations as a matter of routine. A person may read yours after a crisis alert (above), or if you ask us to look at something. While My Sign is in private testing, a member of the team can also read conversations to check the app is behaving. That access stops when the pilot ends.
Notification services. If you turn on notifications, the morning message is delivered through your device's push service: Apple, Google, Mozilla or Microsoft, depending on your device and browser. The message is encrypted end to end, so they cannot read it. They do see that a notification was delivered.
Google Fonts. The app's typefaces load from Google's font servers, which receive your IP address and browser details when a page loads. They do not see anything you write.
When you email us, the message is stored by AWS for 90 days and forwarded to the team's inbox (currently Gmail), where it is kept for as long as we need it to help you.
Nobody else. No advertising, no data brokers, no analytics service that can see what you write. We do not sell data and we do not share it with anyone not listed above, except where the law requires us to.
Other people's details
When you compare charts with someone, you give us their name, how they relate to you, and their birth date, place and, if you know it, time. We use it only to compute the comparison for you. Their computed compatibility and their name may be sent to Anthropic when you ask about them, but their raw birth details are not. We never contact them, and they never become a user.
Please only add someone if you have reason to think they would be comfortable with it. Removing a person deletes their birth details and their compatibility report. Their name will still appear in your conversations and memories until you delete your account. If you are someone who has been added and want your details removed, write to support@mysignapp.com.
What our own logs contain
Our operational logs record the shape of what happens, and never the content. A log line says that a message was sent, how long it was, and whether it was routine or urgent. Log lines identify you only by an opaque account ID. The logging layer is built to replace message text, email addresses and birth details with a character count, and an automated test fails if a message body reaches a log line.
How long we keep it
While your account exists: everything listed above, so that memory works. That includes your conversations, check-ins and daily memories, with no expiry.
Sign-in codes: they work for 15 minutes and only once. The record of the code is erased within about an hour, and the database itself may take up to 48 hours more to sweep it away.
Short-lived bookkeeping: message counts and the day's line are kept for a few days.
Support emails: 90 days in our mail storage, plus the forwarded copy in the team's inbox.
Anthropic: within 30 days, or up to 2 years if flagged for a usage-policy violation (above).
When you delete your account: every record is removed from our database immediately. That covers your chart, conversations, check-ins, memories, the people you added, your notification subscription and your sign-in details.
After deletion: encrypted database backups exist for disaster recovery and are retained for up to 35 days. Nothing is restored from them except to recover from a failure affecting everyone, and never to recover an individual account you asked us to delete. Logs (which hold your account ID but not your words) age out on the 90-day schedule above. Anything already sent to Anthropic or PagerDuty follows their retention, above.
Your controls
Take your data. Settings → Take your data downloads, as a file, your account details, chart, conversations, check-ins, memories, people and everything else we hold under your account. Notification keys are left out because they are device secrets, and so are values we can recompute from your chart.
Delete everything. Settings → Delete everything. Immediate, and it asks you twice. There is no way yet to delete a single message or a single day. Account deletion is the control.
Correct things. Add or change your birth time, your country and your astrology tradition in Settings, and remove anyone you added. For anything else, write to us.
Turn notifications off. Settings, or your phone's own controls.
Ask a person.support@mysignapp.com (until it is live: somi.safavi@gmail.com). You can ask what we hold, ask us to correct or delete it, or object to how we use it. Depending on where you live, you may also have the right to complain to your data-protection authority.
Age
My Sign is for people 18 and over. We check your birth date at signup and refuse accounts below that age.
What My Sign is not
My Sign is for reflection, not prediction. It is not a medical, legal, or financial service, and it is not a crisis service. If a conversation suggests you may be in danger, the app stops and gives you a real crisis line for your country. If you are in immediate danger, contact local emergency services.
Third-party data in the app
Planetary positions are computed with the Swiss Ephemeris by Astrodienst AG, used under licence. The calculation runs on our servers; nothing is sent to Astrodienst.
Birthplace lookup uses data from GeoNames (geonames.org), licensed CC BY 4.0. It is bundled with the app; nothing is sent to GeoNames.
Changes
If this page changes in a way that affects you, we will tell you in the app before it takes effect. If we ever want to send your data somewhere new, such as a different AI provider, we will ask you first.