Privacy Policy

⚠️ Founder-reviewed interim version — posted 2026-11-01. This page has been reviewed by BNC Solutions LLC's founder for accuracy about how Mimi Care actually operates. An independent attorney review is in progress. We will update this page and note the date when that review completes. Material changes require your re-acceptance per Section 13A of the Terms of Service.

Compared with version 2.19: Corrections, and one addition. The addition: on our website's Help library, at help.mimicare.ai, we keep "Did this help?" answers and the words of searches that find nothing, linked to no one, for about 30 days (§2, §8). The corrections: Mimi Care reads steps and falls only on an iPhone (steps, and falls an Apple Watch records), never on Android (§2, §8G). A family Account Holder can set the Primary User's sharing choices only until the Primary User's phone is first linked; after that, only the Primary User can (§8F). The list of what the instant copy of your data holds is now complete (§9).

Mimi Care Privacy Policy

Effective date: 2026-11-01 Last updated: 2026-11-01 Version: 2.20

1. Who we are

Mimi Care is published by BNC Solutions LLC ("BNC", "we", "us"). Mimi Care is a voice-first mobile app that helps people understand photos of bills, mail, medication labels, screens, and possible scams by analyzing them with AI, and that the Primary User can simply talk with.

• Website: https://mimicare.ai
• Contact / privacy requests: support@mimicare.ai

This policy covers the Mimi Care mobile app on iOS and Android, our backend API (served at mimicare.ai/api/*), and the mimicare.ai website (including subdomains app.mimicare.ai, admin.mimicare.ai, platform.mimicare.ai, and help.mimicare.ai).

In this policy, the "trusted contact" (also the "Account Holder") is the adult who sets up and pays for the account; the "Primary User" is the person whose device is linked and who uses the app day to day. These are the same defined terms used in our Terms of Service. "Family on the account" means the Account Holder and anyone else the account adds to help the Primary User.

2. What we collect

We collect what we need to run the service. Several features below are optional; we collect their data only when they are used.

Account data (from the trusted contact who sets up an account)

• Email address
• Phone number (optional — used for the Primary User's "Call for Help" button, not for account contact)
• Apple, Google, or email magic-link sign-in identifier
• Subscription plan and renewal status

Primary-user setup data (entered by the trusted contact or the Primary User)

• Primary User's first name (required)
• Primary User's full legal name (optional — used only to help the AI recognize whether mail is addressed to them)
• Household member names and relationships, up to 5 (optional — same purpose)
• Help contacts — the people the Primary User can reach with "Call for Help": a name or label and a phone number for each, and optionally the days and hours each one should be called
• Preferred language (English or Spanish) and time zone

Helps (when the Primary User asks Mimi Care to explain something)

Every help is stored on our servers, encrypted, by default (see §8E for how it is protected and how long it is kept):

• The photos the app sent. The app normally reduces each photo to at most 1,600 pixels on its longest side before sending it (Show Me camera frames to 1,024 pixels), and what it sends is what we store. If the phone cannot reduce a photo, the app sends the original instead.
• A small preview image (thumbnail) of the help
• The question the Primary User asked, typed or spoken (a spoken question is kept as text)
• The AI's explanation, summary, and any flags, and every follow-up question and answer
• The spoken version of the explanation (the audio the app plays)
• For a Show Me session (live help with the camera), the conversation as text and the camera frames

Scan metadata (created with each scan)

• Category (bill, medication, mail, scam, tech, home, other)
• Timestamp and page count
• Two flags produced by the AI: flagged_scam and flagged_unusual, and a short flag reason if one of them fires
• Whether the scan was answered or refused, and why it was refused
• How many follow-up questions were asked, and how long the answer took
• A fingerprint (SHA-256 hash) of the first photo, used to match a report of a known illegal image to the help it came from (§8A)

Health information (optional)

Only on an iPhone, and only if the Primary User turns on activity or fall tracking and allows access in Apple Health: daily step counts, and falls that an Apple Watch records in Apple Health (iOS 18 and later). Mimi Care does not read health data on Android. See §8G.

Medicines (optional)

Medicines the family or the Primary User add: medicine records, the schedule, doses taken or missed, the name to use on medicine records, how long a course of medicine lasts, and, if a family member adds them, a recorded voice clip and a photo of the pill. Adding a medicine by photographing its label sends the photo to the AI to read the label; we keep the medicine details, not that photo. See §8H.

Reminders and calendar (optional)

Reminders and calendar events the Primary User, the family, or an approved email sender (§8B) adds. See §8I.

Email sent to the Primary User's Mimi Care address (optional)

Messages that approved senders send to the Primary User's own Mimi Care email address, and the senders the family approves. See §8B.

Conversation recordings (optional)

Recordings the Primary User chooses to make with the conversation recorder, and their transcripts, summaries, and questions and answers about them. See §8D.

Subscription records

• Plan tier (shown in-app and at mimicare.ai/pricing at time of purchase)
• Receipt identifiers from Apple, Google, or Stripe
• Start, renewal, and cancellation dates
• Plan changes: the plan you moved from and to, when the change took effect, and changes that were scheduled or undone

Device data

• Apple (APNs) or Google (FCM) push token, so we can send notifications (see §5 for what we send)
• App version, OS version (for crash triage)
• A locally generated device ID
• The phone's model, and the name the phone gives itself (for example, "Mom's iPhone"), for every phone signed in to Mimi Care. A family member can also give the Primary User's phone a nickname.
• On the Primary User's phone, a snapshot of its display settings, sent when the phone is set up and updated when the app starts while Mimi Care's notifications are allowed: the text size, the screen size and pixel density (which show whether the display is zoomed in), and whether Bold Text and Reduce Motion are on. We use it to see how the app looks on that phone when we fix problems or improve the app.
• On the Primary User's phone, whether Mimi Care's notifications are allowed (allowed, turned off, allowed only quietly, or not yet asked) and, on Android, whether on-time alarms ("Alarms & reminders") are allowed. The app sends this when it opens and whenever it changes, and we keep only the latest answer for each phone. If notifications are turned off, family on the account see that on the Primary User's card (§6).
• We keep these device details until the account is deleted — or, for the Primary User's phones, until their data is deleted 60 days after a reported death or 30 days after they are removed from the account. A phone that is replaced or signed out stops getting notifications, but its details are kept until then, and the Primary User's earlier phones stay in the phone history family can see (§6).
• When the app checks for an update: the device ID, app version, and the country the request came from, kept for 90 days

Approximate location from your internet connection

Mimi Care never uses the phone's GPS and never asks for permission to use your location. Like any online service, our servers see the network (IP) address each request comes from, and Cloudflare, our infrastructure provider, estimates from it roughly where that connection is — usually a town or city, sometimes only a region. We use that estimate in three ways:

• Weather. When the Primary User asks Mimi Care about the weather without naming a place, we use the estimate (approximate coordinates, town or city, state, and time zone) to get a forecast for their area. If they name a U.S. town or city instead, we use that place. We round the coordinates to two decimal places — an area about 1 km (two-thirds of a mile) across — and request the forecast from MET Norway (see §6). The request comes from our servers, so MET Norway never receives the Primary User's network address or anything else about who asked. The name of the town or area is given to the AI (§5) so Mimi Care can say it. We do not store the location: the forecast is kept for a few hours at most in Cloudflare's cache, labeled only by those rounded coordinates, so people asking about the same area share one forecast.
• Where a law asks for an age check. For the Primary User's requests, we check the state (or, outside the US, the country) of the estimate against our list of places whose law asks apps like Mimi Care to offer an age check — today only New York — to know whether to offer the check and whether a few conversation features stay off until it is done (see "One-time age check", below). We use only the state or country for this, never the town or coordinates. If we turn on recording of those places, we also keep, for each such place a Primary User's requests came from, the first and latest time, updated at most every 6 hours; places with no such law are never recorded.
• US-only service. We check which country each request comes from, because Mimi Care is offered only in the United States (§7). If a request comes from outside the US, we block it. For each blocked request we keep the country, the requester's network (not the full IP address), the browser or app description (user agent), and the page or part of the service requested, for about 3 months, to investigate access from outside the US. These records are not linked to an account. We also save the country a request came from with app-update checks (above), and when someone sends a message through the contact form on mimicare.ai or joins our waitlist, and with our own measurement of ads and links ("Visits to our website", below).

One-time age check (only where a law requires it)

Some places' laws ask apps like Mimi Care to offer an age check before certain conversation features are on. Today that is only New York (New York General Business Law Article 48), starting January 1, 2027. Where such a law applies, Mimi Care asks the Primary User — never a family member —:

• Where they are. "Are you in New York right now?" We keep the answer (yes or no) and when it was given. After a "no", we do not ask again about that place for 30 days.
• An optional check, two ways. The check starts only after the Primary User agrees on its screen ("I agree — start"); we keep when they agreed and which version of the words they saw, and pass that to Didit. The usual way: the phone's camera takes a picture of the Primary User's face, checks that it is a live person, and estimates their age. If they prefer, they can use a photo ID instead: the phone takes a photo of the ID and a picture of their face, Didit reads the date of birth on the ID, checks that it is a live person, and compares the face with the photo on the ID, only to confirm the ID is theirs. The pictures are used only to check age, never to identify anyone or to search for anyone. They go straight from the phone to our age-check provider, Didit (§6), which processes them in the European Union (Ireland). They never reach our servers or any AI model, and we delete them at Didit as soon as the result reaches us; if a check is started but never finished, we delete it at Didit within about a day. Didit is set not to keep faces for searching, and records no audio. A face estimate of 25 or older, or an ID showing 18 or older, passes; under 18 shows someone under 18; anything else is not a result, and the check can be tried again, either way. If both ways were completed and neither could tell, we treat the Primary User as an adult, unless a check showed signs of someone trying to get around it (for example, a picture of a picture, or a face that does not match the ID).
• What we keep. The result: whether the check showed an adult, showed someone under 18, or could not tell; how it was done (a face estimate, a photo ID, a review, or the "treated as an adult" rule); the lowest age it proved (18 or 25); the place; and when. We never receive or keep the pictures, the estimated age, the date of birth, or anything else from the ID. We also keep a record of each check started — the agreement (when, and which words), when it started, finished, and was deleted at Didit — so we can show the agreement and limit how many checks can be started.
• The five-year age-check record. As New York's rules require, we keep for five years a record that an age check happened, how, when and its result, including that a check was started, that it showed signs of someone trying to get around it, and that a review was asked for and decided. It holds a reference number for the Primary User's seat, the place, the method, the result and the time — never a name, a picture, the estimated age, the date of birth or any document. It is kept for the five years even if the account is deleted, and then deleted.
• Skipping it. The check is optional. Until the Primary User passes it, in a place whose law asks for it, Mimi Care still helps them but does not use or add to what it remembers from earlier conversations, does not bring up things they have shared, and talks more plainly (§8F). What Mimi Care already remembers is kept, and comes back once the check is passed.
• Under 18, and asking for a review. If a check shows the Primary User is under 18, those features stay off on their phone, and we email the Account Holder. The Primary User can try again from the home screen, including with a photo ID. The Account Holder can also ask us to review the result on our website, and can do the same when checks could not tell the Primary User's age: they send one document that shows the Primary User's age (a photo ID is welcome but never required) and, if they want, a short note. The document and the note are stored encrypted, only our staff who decide reviews can open them, and each opening is recorded in our audit log. We email the Account Holder the written decision, and we delete the document and the note as soon as we decide.

How the app is used, feedback, and messages to us

• App-start and speed measurements — that the app was opened, how long it took to start, and how long Mimi Care took to answer. These are kept in Cloudflare Analytics Engine for about 3 months, labeled with a one-way coded (hashed) seat reference rather than the account, and then expire on their own; they cannot be deleted one by one.
• Other product usage events — for example, that a screen such as the subscription screen was shown — linked to the seat and the account. We use them to understand how the app is used and what is not working.
• Feedback and surveys — feedback given in the app or to support, and answers to short surveys we may send.
• "Did this help?" answers and Help searches on our website — in the Help library on our website (help.mimicare.ai), when you answer "Did this help?" under an article, we keep which article it was, your yes or no, the page's language, and the time. When a search in the library finds nothing, we keep only its words, simplified: in lowercase and without accents, without common words such as "how", "my", or "the", without any word that contains a number, and no more than 8 words. We never keep the search as you typed it, and a search that finds articles sends us nothing. These records are not linked to anyone: they include no account, device, network (IP) address, browser description, or cookie. We use your network address only for a moment, to limit how many answers one connection can send, and do not keep it with them. We use these records to learn which articles help and what to write next, and delete them about 30 days later.
• Messages to us — messages sent through the contact form on mimicare.ai or to support@mimicare.ai, and waitlist sign-ups: your name, email address, and message, and details such as how you heard about us, the page you came from (with any ad click ID replaced by a placeholder), your browser's user agent, and the country of the request.

Joining the waitlist by signing up

While Mimi Care is letting people in a few at a time, if you sign up in the Mimi Care app or at app.mimicare.ai with Apple, Google, or your email address before you are invited, you join our waitlist instead of getting an account. We keep:

• The same details as a waitlist sign-up on our website (above): your email address, the language you use, your browser or app description (user agent), the country of the request, and your answers to our questions — who you are setting Mimi Care up for, how you heard about us, and what you hope it helps with. If you already joined with the form on our website, we keep your earlier answers.
• If you signed up with Apple or Google, the sign-in identifier it gives us and the name it shares with us, if any, so that your account can be created with them once you are invited.
• Whether you signed up on an iPhone, an Android phone, or the web.
• If you signed up at app.mimicare.ai after following one of our ads or links, the page you arrived on and the campaign tags in that link (such as utm_campaign), so we can tell which of our ads and links bring people to Mimi Care (see "Visits to our website" below).
• A waitlist pass: a random code that lets the app or website show your place on the list and, once you are invited, sign you in once without asking you to sign in again. In the app it is kept in the phone's secure storage (on iPhone, the Keychain, which can keep it after the app is deleted); on the web it is kept in a cookie that only our servers can read. We keep only a one-way coded (hashed) form of the code, never the code itself. A pass stops working when it is used to sign in, when you sign in to Mimi Care with the same email address any other way (for example, on another phone), when you choose "Use a different account", when you leave the waitlist, or about 90 days after it was last used. We keep at most five working passes for one email address; signing up on a sixth phone or browser stops the pass that was used least recently.

You can remove yourself from the waitlist on the waitlist screen in the app or on the waitlist page at app.mimicare.ai ("Remove me from the waitlist"), or by emailing support@mimicare.ai. Leaving deletes your waitlist sign-up and stops all of your passes.

Visits to our website

On the public pages of mimicare.ai and help.mimicare.ai, and when you sign up at app.mimicare.ai, we measure visits in the ways below. None of this happens in the app, or on the pages you see after signing in.

• Google Analytics. Google Analytics counts visits and shows us which pages are read. For each visit it records the page, the page you came from, your browser, operating system and type of device, your approximate area (which Google estimates from your network address), and actions such as scrolling down a page, following a link, or joining the waitlist, under a random identifier kept in a cookie in your browser. We turn off Google signals and ad personalization, so what it records is not used to show you ads. Before a page address is sent to Google, we replace any ad click ID in it (for example, fbclid) with a placeholder, and if you go on from a public page to a signed-in one, Google is sent a placeholder instead of that page's address.
• Our own measurement of ads and links. On the pages our ads and links lead to (mimicare.ai/start and the pages under it), we record that the page was viewed, which buttons were pressed, and whether the video was played. When someone joins the waitlist or creates an account at mimicare.ai or app.mimicare.ai, we also record that a sign-up happened; signing up in the Mimi Care app is not recorded here. Each record holds the campaign tags in the link you arrived from (such as utm_campaign), which ad platform the link came from, the name of the site you came from, your country, whether you are on a phone, tablet, or computer, and the language of the page. The records from one visit share a random visit number that your browser keeps only until you close the tab. We do not store your network address, your browser's description, the ad platform's click ID, or your name or email address in these records.
• Which ads and links lead to accounts and purchases. When you sign up at mimicare.ai or app.mimicare.ai after following one of our ads or links, we keep that link's campaign tags and the ad platform it came from with your waitlist sign-up and, once you have one, with your account. That lets us count, for each ad or link, how many sign-ups were let in from the waitlist, became accounts, and bought a pass or a plan. We don't keep your network address or the ad platform's click ID with them, we never share them, and they are deleted after 13 months, or sooner when you leave the waitlist or your account is deleted.
• Short links. Our posts and printed materials use short links such as mimicare.ai/s/12. When one is opened, we add one to that link's count for the day and send you on to the page it points to. We record nothing about who opened it.

Companion conversations (optional)

Only when the Primary User turns on "Let Mimi Care remember our conversations":

• Short notes on what each conversation was about, and a few topic words — written by the AI after the conversation ends
• Things Mimi Care remembers to make later conversations more personal (for example, "her grandson is named Max"), including notes a family member adds
• The time and length of each conversation

Our app does not store the audio of these conversations. A text record of what was said and Mimi Care's replies is kept in the record of AI conversations (§8J). Whether or not remembering is on, conversations can produce the wellbeing signals and safety alerts described in §8F.

Notice at collection (California). We collect the categories above to provide and secure the service, process payments, and comply with law, as detailed in §5. We retain each category for the period described in §8. We do not sell or "share" personal information and do not use it for cross-context behavioral advertising. For the statutory categories, sources, and recipients, see the table in §9.

3. What we do not collect or store, and what is encrypted

• Photos. We store the copy of each photo the app sends, encrypted, as part of the help (§8E). The app normally sends a reduced copy, not the camera's full-resolution original (§2).
• Age-check pictures. The face picture, and the photo of an ID if the Primary User chooses that way, taken for the one-time age check go only to Didit, which uses them only to check age and deletes them right after the result (§2). We never receive or store them, the estimated age, or the date of birth. A document the Account Holder sends us for a review of an under-18 or unclear result is stored encrypted and deleted as soon as we decide (§2).
• What is encrypted. Helps, conversation recordings, Companion notes and memories, daily activity, medicine details, reminder and calendar text (title, notes, and location), and the subject and body of email delivered to the Primary User's Mimi Care address are stored with an extra layer of encryption (§8E, §11). The record of AI conversations is encrypted separately, to a key our servers do not have (§8J).
• What is not. Some other records are stored without that extra layer, protected by the security of our database and file storage: blocked (quarantined) email, the short notes in wellbeing signals, fall events, dose records, help contacts, scan flags and flag reasons, and the sender's address and display name of email to the Primary User's Mimi Care address. Our AI request logs (§5) hold none of what is sent to the AI or its answers, only a record that each request happened.
• Spoken questions. We do not keep the audio of spoken questions. When the Primary User asks a question out loud, the audio is turned into text and the audio is then deleted. The text is checked by our safety classifiers and is kept as part of the help (§8E), or as part of a safety-incident record if it triggers one (§8A), and in the record of AI conversations (§8J). The audio we do keep is: conversation recordings the Primary User chooses to make (§8D), the spoken version of each help's explanation (§8E), and medicine voice clips a family member records (§8H).
• Conversations with Mimi Care. Our app does not store their audio. A text record of what was said and Mimi Care's replies is kept in the record of AI conversations (§8J). Beyond that, only the short notes described in §8F are kept, and only if the Primary User turns remembering on.
• Contacts. We do not read the phone's address book. The only contacts we store are the help contacts entered in the app (§2).
• Location and tracking. We do not collect precise (GPS) location, your browsing history on other sites, or advertising identifiers — when you arrive from an ad, we keep only which ad platform the link came from, never its click ID (§2) — and the app never asks for permission to use your location. The only location we use is the approximate area of your internet connection, as described in §2.
• We do not sell personal information. Ever.

4. What the app keeps on the Primary User's device

The app keeps some items on the Primary User's phone so it works quickly and offline:

• Device ID, seat ID, first name, and the family phone number, in the phone's secure storage; help contacts, in the app's own storage
• A local copy of past helps (category, timestamps, flags, and the full AI explanation text), and copies of their photos and spoken explanations — the first place the app looks when replaying a past help
• Voice, language, and font-size preferences

Most of this also exists on our servers: the seat ID, first name, and help contacts are part of the account (§2), and every help is stored on our servers (§8E). Uninstalling the app removes most of the copy on the phone. On iPhone, a few items (such as the device and seat IDs, the first name, and the family phone number) are kept in the phone's Keychain, which can remain after the app is deleted and is included in the phone's iCloud Keychain. Uninstalling does not delete anything on our servers — delete helps in the app, or ask us to delete data as described in §9.

5. How we use the data

• Run the app: explain what the Primary User photographs, answer their questions, and let Mimi Care talk with them
• Send notifications: reminders and medicine-dose reminders to the Primary User; and, to family on the account, safety alerts and SOS alerts (§8F), possible-scam alerts, missed reminders and medicines, requests for approval, new items in the family Inbox, and notes about new versions of the app
• Let the family on the account know when the Primary User may need them (§8F) — by push notification, or by email to the Account Holder when a notification cannot reach anyone
• Keep the short conversation notes and memories the Primary User chose to let Mimi Care keep (§8F)
• Keep the record of AI conversations (§8J), so that we can show what Mimi Care was asked and what it answered if that is ever needed for a legal claim or a legal obligation
• Run the optional features the account uses: health tracking (§8G), medicines (§8H), reminders and calendar (§8I), the Mimi Care email address (§8B), and the conversation recorder (§8D)
• Process subscription payments through Apple, Google, or Stripe
• Answer weather questions for the Primary User's area, keep the service US-only, and know whether a place's law asks for an age check, using the approximate location described in §2
• Offer the one-time age check where a law requires it, keep a few conversation features off until it is passed, email the Account Holder if a check shows the Primary User is under 18, decide a review they ask for, and keep the five-year age-check record the law requires (§2)
• Detect abuse and safety risks, and rate-limit excessive requests
• Debug crashes and errors, and understand how the app is used so we can improve it
• Measure visits to our website and learn which ads and links bring people to Mimi Care (§2, "Visits to our website") — never to show you ads
• Help with support requests (see "Staff access" in §6)
• Comply with law

We do not use your data to train AI models.

How the AI works, and what Anthropic receives. The AI that reads pictures, answers questions, and writes Mimi Care's side of a conversation is Claude, made by Anthropic, PBC. To produce each answer, Mimi Care sends Anthropic the content that answer needs — the photo, the question, or the conversation so far — through Anthropic's API, routed via Cloudflare's AI Gateway. In addition:

• On every turn of a conversation with Mimi Care, we send Claude the day's medicines, reminders, and calendar, so it can answer questions about the day, and — if remembering is on — what Mimi Care remembers and recent conversation notes.
• When the Primary User asks about past helps or recordings, Mimi Care sends what is needed to answer.
• When the Primary User asks Mimi Care to call someone, the names (labels) of their help contacts are sent so the right one can be found; phone numbers are not.
• Messages sent to the Primary User's Mimi Care email address are sent to Claude to check for scams and to find reminders and appointments (§8B).

Anthropic's Commercial Terms prohibit it from training models on this content. Under Anthropic's published data-retention policy, it deletes API inputs and outputs from its systems within 30 days, and keeps them longer (up to 2 years) only if its trust-and-safety systems flag them for review. Mimi Care can alternatively run Claude through Amazon Web Services (AWS Bedrock), whose terms likewise prohibit training on inputs or outputs.

Our AI request logs. Every request Mimi Care sends to Claude, and Claude's answer, passes through Cloudflare's AI Gateway in Mimi Care's own Cloudflare account. Since 2026-10-13 the gateway keeps only a record that the request happened: the time, the AI model, the size of the request and answer, the cost, how long it took, whether it succeeded, and reference codes for the account, the person, and the feature used instead of names. It does not keep the photo, the question, the conversation, the information sent with it, or the AI's answer; what it kept before that date has been deleted. The backup transcription of conversation recordings (§8D) passes through the same gateway on the same terms. We use these records to track costs and fix problems. There is no fixed time limit: the gateway keeps up to a set number of records and deletes the oldest when that limit is reached. Deleting a help, a recording, a conversation, or an account does not remove these records, but they hold none of that content. What is sent to the AI and its answer are kept only in the record of AI conversations (§8J).

Follow-up questions and the 5-minute cache. To make follow-up questions about the same document fast, parts of a request may be held in Anthropic's prompt cache for 5 minutes after they were last used, then discarded.

Web lookups. For some questions, Claude may search the public web for current information (for example, a company's published phone number). The search terms are written by the AI from the question being answered, and Anthropic carries out the search, which can involve its own search provider.

Automatic safety checks. Every photo, and what the Primary User says or types, is also checked by small automatic safety classifiers that run on Cloudflare (Workers AI). These checks can refuse a request — for example, sexually explicit photos are refused and not saved — or create a safety incident (§8A).

When a photo is checked. In the newest version of the app, the safety check on a photo can start as soon as the Primary User taps Done in the camera, before they ask their question and send the photos, so the answer comes sooner. This includes photos they then do not send, or take again. The photo is sent to Cloudflare only for that check and is not kept. We keep the photo's fingerprint (a SHA-256 hash) and the check's result for up to 15 minutes, so the help they send next can use that result once instead of checking the same photo again; it is deleted once used, or after that time. When a check finds a sexually explicit photo, a weapon, or violence, we keep the fingerprint and the check's safety scores for 24 hours, whether or not the photo was sent, so the same photo gets the same answer. Each check is kept in the record of AI conversations like every other check (§8J).

6. Who we share it with

We share only with the subprocessors we need to run the service. A current list is at mimicare.ai/subprocessors and in docs/legal/subprocessors.md. Today that includes Anthropic (Claude, the AI model), Amazon Web Services (S3 for disaster-recovery backups; Bedrock as an alternative way to run Claude), Cloudflare (infrastructure, including its AI Gateway and the AI request logs it keeps for us, Workers AI safety checks, Analytics Engine, and email delivery), Deepgram (speech-to-text and the app's voice), OpenAI (backup speech-to-text), Twilio (text-message alerts to our staff), Stripe (web billing), Apple (App Store billing and APNs push), Google (Play Billing, FCM push, and Google Analytics on our public website), Sentry (error reporting), and Didit (the one-time age check where a law requires one, §2).

Family on the account. Family members always see how often the Primary User uses Mimi Care (for example, when they last used it and how many helps they asked for), and that a recording was made, when, and how long it was. They also see the Primary User's phone history — each phone's model or name, its system version, any nickname a family member gave it, and when it was set up or replaced — and, if notifications are turned off on the Primary User's phone, "Notifications are off on [name]'s phone." That notice adds "so medicine reminders can't sound" only when the Primary User shares medicines with that family member and has medicine reminders on. Beyond that, they see what the Primary User's sharing choices allow (§8D–§8I), and they receive the safety alerts described in §8F.

Staff access. Authorized Mimi Care staff can, to help with support, view an account the way its family sees it; during that session they see whatever the family can see, and each session is audit-logged. Our administrative tool for protected information can also open, with a recorded reason and an audit-log entry, the Primary User's full legal name and household names, email delivered to the Primary User's Mimi Care address (§8B), safety-incident content (§8A), and legal-consent records. Staff who decide a review of an under-18 age-check result can open the document and note the Account Holder sent for it, and each opening is audit-logged (§2). The record of AI conversations can be opened only as described in §8J.

Weather forecasts. When the Primary User asks about the weather, we request a forecast from MET Norway (the Norwegian Meteorological Institute, api.met.no), a public weather service. It receives only rounded coordinates (§2) — never a name, account, network address, or anything else about who asked — and returns a forecast anyone can request, so it is a public data source rather than a subprocessor. We list it on the subprocessors page under "Public data sources."

Authorities. We share with law enforcement under valid legal process. In addition, as described in §8A, apparent child sexual abuse material is reported to the National Center for Missing and Exploited Children (NCMEC) as federal law requires, and our staff may decide to report a credible threat against another person to the FBI or local law enforcement, to give identifying information to local law enforcement for a welfare check when someone appears to be at imminent risk of harming themselves, or to report suspected elder abuse to Adult Protective Services. We do not sell data or share it for cross-context behavioral advertising.

7. Where data is stored

Backend infrastructure (Cloudflare Workers, D1, KV, R2) operates in the United States. Subprocessors process data in the United States, except that Google may process Google Analytics data about visits to our website in other countries where it operates (§2), and Didit processes the pictures taken for the one-time age check in the European Union (Ireland) and deletes them right after the result (§2). Weather requests go to MET Norway in Norway and carry only rounded coordinates (§2).

US-only service. Mimi Care is offered only to residents of the United States. We do not accept users from outside the US. We check where each request comes from, and requests from outside the US are blocked (§2).

8. Retention

Safety records — wellbeing signals, safety incidents, and referral records — survive account deletion: when an account is deleted, they are detached from the Primary User's seat and kept for the periods above, then deleted. A legal hold on an account pauses its deletion.

Backups. A nightly copy of our database is kept in Amazon S3 for disaster recovery. Data that has been deleted from the service — including records described above as deleted — can remain in those copies until they expire under our backup schedule. Records stored encrypted stay encrypted in the copies.

8A. Safety-incident retention exception

• When our automated classifiers detect a credible threat of violence against another person, or images of weapons or violence; imminent self-harm indicators; or elder-abuse indicators — including signs that someone else on the account is trying to misuse Mimi Care to pressure or mislead the Primary User — we create an access-controlled safety incident record. The triggering content (the words, for voice and text; an image fingerprint and classifier output, for images) is preserved in this record, together with an identity record (see §8), so that authorities can be contacted and can subsequently identify the submitter. When the trigger is something a family member wrote to Mimi Care, that family member's words and identity record are what is preserved.
• We also create a safety incident when a Primary User answers, on the setup screen, that they do not know the person who set up their phone. No classifier is involved and no words are preserved; our staff are alerted so they can look into it.
• Our staff are alerted by email, and for the most serious incidents also by text message. These alerts carry a short description of the kind of incident and internal reference numbers, not what was said.
• Incident content is encrypted at rest, accessed only by named officers and counsel under two-factor authentication, and purged automatically at the retention boundary in §8 unless a legal hold requires otherwise.
• People decide what happens next. Our staff decide whether to report a credible threat to the FBI or local law enforcement, to give identifying information to local law enforcement for a welfare check when someone appears to be at imminent risk of harming themselves, or to report suspected elder abuse to Adult Protective Services; nothing is referred to them automatically.
• CSAM. We rely on Cloudflare's CSAM Scanning Tool, which may report hash matches of known child sexual abuse material to NCMEC. If we are notified of a match for a help we stored, we create a safety incident, keep an unaltered, still-encrypted copy of that help for 7 years as evidence, separate from the Primary User's helps, and record the NCMEC report number. Sexually explicit photos are refused by our safety checks and not saved.
• Political content is never preserved. Our classifier filters politically-referenced content to a plain refusal; no incident is created; no authority is contacted.
• You may request a list of any incident records associated with your account by emailing support@mimicare.ai, except that CSAM incident existence may not be disclosable in certain circumstances under federal law.

8B. The Primary User's Mimi Care email address

How it works. Each Primary User can have their own Mimi Care email address. The family approves who may send to it; each approved sender confirms by replying once from their address. Messages from approved senders that pass their email service's authentication checks are read by Claude (§5), which checks each message for signs of a scam and finds reminders and appointments to add to the Primary User's reminders or calendar (§8I).

What we keep. Each message's subject and body, stored encrypted, together with its sender's address and display name (not encrypted), for 90 days, then deleted automatically. Authorized Mimi Care staff can open a message's subject and body in our administrative tools when needed to support the account or investigate abuse; each such access requires a stated reason and is audit-logged.

Possible scams. A message that looks like a scam is never added to reminders or the calendar. The family is told in their Mimi Care Inbox, with a push notification, as the Primary User's sharing choices and the family's notification settings allow.

Blocked email (quarantine). Emails that are blocked before reaching the Primary User — due to authentication failures (DMARC / SPF / DKIM), an unrecognized recipient address, an unapproved sender, a rate-limit excess, a disallowed attachment type, or a malformed calendar invite — are retained in a quarantine store for up to 90 days.

• This content was never delivered to the Primary User and is not part of the Primary User's Mimi Care experience.
• It is stored solely to allow Mimi Care staff to investigate spoofing campaigns, abuse patterns, and configuration errors.
• The retained content may include the sender's email address, display name, subject line, and message body (plain text only, capped at 4,000 characters). It is not stored with the extra layer of encryption described in §3.
• Quarantine records are accessible only to Mimi Care staff under two-factor authentication and are automatically purged after 90 days.
• No AI model processes quarantine content; it is stored and reviewed by humans only.

8D. Conversation recordings (optional feature)

When the Primary User uses the conversation recorder, Mimi Care:

1. Records audio — Only when the Primary User actively taps "Record a conversation" on their device. The app shows a consent card the Primary User can show the other person before tapping "They said yes — start." Recording cannot start without two taps.

2. Transcribes — Audio is sent to Deepgram for transcription. If transcription is unclear, we fall back to OpenAI Whisper via our Cloudflare AI Gateway. We ask Deepgram not to use our requests to improve its models.

3. Summarizes and answers questions — The transcript is sent to Claude through Anthropic's API (see §5) to generate a summary and key items, and to answer questions the Primary User or family ask about the recording.

4. Keeps the recording — The audio, transcript, summary, and questions and answers are kept for up to 180 days from the recording, then deleted automatically, unless the Primary User — or a family member who can read the recording — marks it Keep forever; a recording marked Keep forever is never deleted automatically (this shares the 50-item limit with helps, §8E). The Primary User can delete a recording, or just its audio, at any time.

5. Encrypts it — The audio, transcript, summary, and questions and answers are stored encrypted at rest, under the same platform key and record binding as helps (§8E). Staff can see a shared recording only when viewing the account as its family sees it for support (§6). The transcript and questions sent to Claude, and Claude's summaries and answers, are also kept as text in the record of AI conversations (§8J), which deleting a recording does not remove.

6. Shares only when the Primary User chooses — Family always sees that a recording was made, when, and how long it was. What it says is private unless the Primary User chooses "Share with family" for it. (Some older recordings instead follow the Primary User's choice about sharing their helps.) Family members on the same account can read shared recordings on app.mimicare.ai and ask questions about them.

We do not use recording content for training or for any purpose other than serving the Primary User.

8E. Your past helps (encrypted scan history)

What we keep. Every time Mimi Care helps with something you photograph, we keep a copy of that help on our servers, encrypted: the photos as the app sent them (normally reduced to at most 1,600 pixels on the longest side; Show Me camera frames 1,024 pixels), a small preview image, your question, the AI's explanation, any follow-up questions and answers, and the spoken version of the explanation. A Show Me session is kept the same way, with its conversation as text. This happens for every help, with no setting to turn it on or off. It lets you see and hear your past helps from any device, or after reinstalling the app.

How it's protected. Helps are encrypted (AES-256-GCM) before they are written to our database and file storage. For consumer accounts, all helps are encrypted under one Mimi Care platform key, which is itself encrypted by a master key kept in our secured secret-management systems, never in the database or file storage. Each encrypted record is also bound to the record it belongs to, so it cannot be moved to another record and opened there; the encryption does not tie a help to your account. Who can open a help is decided by our sign-in and permission checks, not by the encryption:

• the Primary User, in the app;
• family on the account, when the Primary User's sharing choices let them see helps (§8F);
• authorized staff, only while viewing the account as its family sees it for support (§6) — so only helps the family can see;
• our backend systems, which decrypt helps to show them and to answer questions about them. Because our backend holds the keys, people who operate it could technically decrypt helps; we limit that access.

A text copy of what was sent to the AI (with each photo replaced by a code that identifies it) and of its answer is also kept in the record of AI conversations (§8J). Our AI request logs (§5) keep only a record that each request happened.

How long we keep it. By default, we keep each help for 180 days from when it was created (or from the last time its "keep" status changed). After that, it is permanently deleted automatically.

Keep forever. If a particular help matters to you, tap Keep forever inside the app; a family member who can see the help can do the same. We will hold onto it until someone taps "Stop keeping" or it is deleted. Up to 50 items can be kept this way, counting helps and conversation recordings together. Tapping "Stop keeping" restarts the 180-day countdown.

Deleting a help. Deleting a help inside the app removes the help — the photos, the preview, the explanation, the conversation, and the spoken explanation — from our servers within minutes. Every deletion is recorded in our audit log. Deleting a help does not remove copies kept under §8A, or copies in our backups until they expire (§8). This does not remove its entry in the record of AI conversations (§8J). Our AI request logs (§5) keep their record that the requests happened, which holds none of the help's content.

What we audit. Every time a help is read, shared, or deleted, we log the action in an access-controlled audit log. This log is used to detect misuse and to comply with data-protection obligations.

What the AI provider receives. A saved help is sent back to the AI when it is needed to answer something the Primary User asks: a follow-up question about the help, or a question to Mimi Care — in conversation or during Show Me — about a past help. The AI provider's handling of the content it receives is described in §5.

8F. Companion conversations with Mimi Care (optional)

The Primary User can talk with Mimi Care about anything. Conversations are processed by Claude as described in §5 (and speech by Deepgram, as described in §6).

What is kept — only if the Primary User turns on "Let Mimi Care remember our conversations." After each conversation, the AI writes a few short notes on what it was about and a few topic words, and may note things worth remembering (for example, a grandchild's name). A family member may also add things for Mimi Care to remember. All of this is encrypted at rest, and each encrypted record is bound to the Primary User's seat, so it cannot be opened as part of anyone else's records. Our app never stores the audio of a conversation. A text record of what was said and Mimi Care's replies is kept in the record of AI conversations (§8J), and words that trigger a safety incident are kept as described in §8A. We keep the time and length of each conversation, without its content, for the life of the account.

What the Primary User controls. The Primary User can see every note and memory in the app's Settings, have a note read aloud, forget any conversation, and remove any memory. When the Primary User forgets a conversation about something sad, Mimi Care keeps only a private reminder of who or what it was about — never what was said. That way, Mimi Care is never the first to bring it up. That reminder is not shown on any screen, and it is deleted when the Primary User turns off "Let Mimi Care remember". Turning it off deletes what Mimi Care remembered and clears every conversation's notes. If the Primary User asks that a topic be set aside, Mimi Care does not bring it up, and notes, memories, and alert text that mention it are withheld — except where a topic concerns safety (for example, money being demanded, a lawyer, or calling for help), which can never be set aside.

What family can see. Family always sees how often the Primary User uses Mimi Care (§6). Of their conversations, only what the Primary User chooses:

• Conversation notes and topic words — only if the Primary User turns on "Share conversation summaries with my family", and only for conversations held while sharing was on. Turning sharing off hides them from family immediately.
• Wellbeing signals — short, plain notes about patterns such as loneliness, never the conversation itself — as allowed by the Primary User's sharing choices.
• Things Mimi Care remembers — family can see what they added and what the Primary User asked it to remember; what Mimi Care noticed only when the Primary User shares summaries.

The Primary User's sharing choices. Beyond conversations, the Primary User's sharing choices decide what family on the account can see of their helps, medicines, reminders, possible scams, how they are feeling, steps, and falls. Each can be turned on or off. Unless someone changes them, reminders, possible scams, how they are feeling, and falls are shared, and helps, medicines, and steps are not. When a family member is the Account Holder, they can set these choices while adding the Primary User, and change them until the Primary User's phone is first linked to Mimi Care. After that, only the Primary User can change them, in the app. Other family on the account can ask the Primary User to share something but cannot change these choices. The Primary User can always see and change every choice in the app, and when they change one on an account a family member manages, the family is told.

"Topics Mimi Care won’t bring up." A family member may ask that a topic — for example, a loss — not be brought up. Mimi Care is then never the first to raise it, but still talks about it whenever the Primary User does. The Primary User can see these requests in the app's Settings. A topic that concerns safety can never be one of them.

"Share our talks with my family." This is on unless the Primary User turns it off. While it is on, family sees what the choices above allow. If the Primary User turns it off — in Settings, or by asking Mimi Care — family sees nothing about their conversations: no conversation notes, no wellbeing signals from them, and none of the things Mimi Care remembers except what family added themselves. Turning it back on never shows family anything from while it was off. Safety alerts still reach family, as described below.

Safety alerts — regardless of sharing settings.

• Possible self-harm, a medical emergency happening now, or someone pressuring the Primary User for money. If something the Primary User says suggests one of these, Mimi Care lets the family on the account know: by push notification to each family member's Mimi Care app, or — if the notification could not reach any family member's app — by email to the Account Holder. These alerts never include the Primary User's words. If the Primary User is also the Account Holder, there is no email, and if no other family member is on the account, there is no one to notify; Mimi Care will still give the Primary User the number to call (988 or 911).
• Possible verbal abuse or coercion by someone the Primary User knows. Mimi Care sends the family on the account a general notification that something may need a caring check-in — without the Primary User's words — unless a family member has turned those notifications off, and alerts our staff (§8A). These detections are not shown on the family dashboard.
• SOS. When the Primary User uses the SOS button, the Account Holder gets a push notification and an email.

Other alerts — falls, missed important medicines, and possible scams found in a help or an email — reach family only as the Primary User's sharing choices allow. Content that triggers a safety incident is preserved as described in §8A.

Why some removed text is kept. A memory a family member added and later removed, or a topic they asked Mimi Care not to bring up and later removed, is kept, access-restricted, until the account is deleted, together with a record of who added and removed it. Mimi Care no longer uses it and it appears on no screen. We keep it because a record of what others asked to have remembered can matter if someone tries to misuse Mimi Care to pressure or mislead the Primary User.

8G. Health information (optional)

If the Primary User turns on activity or fall tracking on an iPhone and allows access in Apple Health, the app reads daily step counts, and falls that an Apple Watch records in Apple Health (iOS 18 and later), and sends them to our servers. Mimi Care does not read health data on Android. Daily activity is stored encrypted; fall events are stored without that extra layer of encryption (§3). Family on the account see steps and falls only as the Primary User's sharing choices allow (§8F). We keep health information for 365 days; if the Primary User turns tracking off, that kind of health data is deleted after 7 days.

8H. Medicines (optional)

Family on the account or the Primary User can add the Primary User's medicines: each medicine's details and schedule, doses taken or missed, the name to use on medicine records, and how long a course of medicine lasts. Medicine details — names, strengths, and pharmacy and prescriber — are stored encrypted; the record of doses taken or missed is stored without that extra layer (§3). Medicine records, including the record of doses, are kept for the life of the account, including medicines that were stopped, so the history stays accurate. A family member can also record a short voice clip and add a photo of the pill, which play or show when a dose is due; these are stored encrypted until a family member removes or replaces them, or the account is deleted. Family can create a medicine report to share with a doctor. Family see the Primary User's medicines only as the Primary User's sharing choices allow (§8F). During conversations with Mimi Care, the day's medicines are sent to Claude on every turn, and more detail is sent when the Primary User asks about a medicine (§5).

8I. Reminders and calendar (optional)

The Primary User, family on the account, or an approved email sender (§8B) can add reminders and calendar events. Their text (title, notes, and location) is stored on our servers with an extra layer of encryption (§3, §11). They are kept for the life of the account; deleting one removes it from every screen, and its record is kept until the account is deleted. Family see reminders as the Primary User's sharing choices allow (§8F), and the Account Holder can be told about a missed reminder if missed-reminder alerts are on. During conversations with Mimi Care, the day's reminders and calendar are sent to Claude on every turn (§5).

8J. Record of AI conversations

We keep a record of every exchange with Mimi Care's AI for the account — whether by the Primary User or by family on the account, or done automatically for them (for example, checking an email sent to the Primary User's Mimi Care address, or summarizing a conversation recording) — so that we can show what Mimi Care was asked and what it answered if that is ever needed for a legal claim or a legal obligation.

• What it contains: the text sent to the AI (the question, or what was said, as text; Mimi Care's instructions to the AI; and information sent with it, such as the day's medicines or reminders), the AI's answer (with any reasoning it wrote and any web pages it cited), what Mimi Care then showed or said, which AI model was used, the time, the app version, and which account and person it was for. It also notes some choices made in the app, such as whether to go on after a safety warning. It never contains audio, photos, or documents — only a code that identifies each photo or document, and its size. Before an account or a person's data is deleted, we add to the record the names, email addresses, and phone numbers kept for them (including an estate contact's), so that their entries can still be found.
• How it is protected: each entry is encrypted, when it is made, to a key our servers do not have; the key is kept offline. Our servers can add entries but cannot read them. (Mimi Care's standard instructions to the AI, which are the same for everyone and contain no personal information, may be stored once without that encryption, and entries point to them.) It is opened only by an authorized officer of BNC Solutions when a legal claim or legal obligation requires it, and every opening is logged.
• How long: 6 years from each entry, then deleted automatically (within about a month after that), unless a legal hold requires us to keep it longer. Deleting a help, a recording, a conversation note, or the account does not remove it, and it is kept after a reported death and when a Primary User is removed. If you ask us to delete your personal information, including your health information (§9), and we verify the request, we erase the account's entries, unless a legal hold requires us to keep them.
• When it starts: only after the account holder accepts version 2.15 of this policy or a later one. Earlier exchanges were not recorded.

9. Your rights

Everyone

• Access / know what personal information we hold about you, and export it (see "How we deliver your data" below)

• Correct inaccurate personal information

• Delete your account and personal information. There are two ways:

  • Close the account from the account settings in the app or on the family dashboard. The deletion is scheduled 30 days ahead; the account keeps working until then, and you can cancel during that time.
  • Ask us to delete your personal information under Your privacy rights at app.mimicare.ai, or by emailing support@mimicare.ai. Our team handles the request within the response time below.

  The Account Holder can also remove a Primary User from the account, from that person's page in the app or on the family dashboard. Their access stops right away, and their personal information is deleted 30 days later, unless the removal is undone first (§8).

  When the account is deleted, a subscription paid by card at app.mimicare.ai or through Google Play — including one that is paused or on hold — stops renewing at the end of the period already paid for, with no refund. Apple lets only the person who bought an App Store subscription cancel it, so we email the Account Holder the steps (Terms of Service, §5).

  Deleted data can remain in our backups until they expire (§8). Closing the account does not remove the record of AI conversations; asking us to delete your personal information does, unless a legal hold applies (§8J).

You can make a request on Your privacy rights at app.mimicare.ai (signed in) or by emailing support@mimicare.ai. We aim to respond within 30 days (extendable once, by up to another 45 days, where the law allows, with notice). Requests to opt out are recorded right away. Before we act, we may ask you to verify your identity using the email or phone on your account; a person on our team does this. You may use an authorized agent to submit a request; we may ask the agent for proof of authorization and may ask you to verify your own identity directly. We will not discriminate against you for exercising any of these rights.

How we deliver your data.

• Instant copy. On Your privacy rights, you can download a copy right away. It holds your account details; your own phones and browsers; the people on your account and their phones; a record of each help (its kind, date, and flags — not the photo or the AI's answer); medicines and each dose; reminders and calendar events, with each time one rang, whether it was done, snoozed, or missed, and what family was last told about it; steps and falls; SOS presses; when each talk with Mimi Care happened and how long it lasted; each Mimi Care email address, with the email sent to it, its allowed senders, and a record of the email it blocked; sharing choices; the settings family chose (such as who the Help button calls), with each ask and each change family was told about; consents; feedback and survey answers; notification settings; the ad campaign you signed up through (when one was recorded); a summary of use; and your plan. For someone else on your account, their helps, medicines, reminders and calendar, steps, and falls are in the file only while they share them with you, and their talks only while they share those with you. Records of how the app was used (for example, which screens were shown) are not in the file.
• Full export. You can also ask our team for a fuller export. When it is ready, we email you. The email does not contain the file or a link to it; it asks you to sign in and open Your privacy rights, where you download the file through a link that works only for you and only for a short time. The file can be downloaded for 30 days, then it is deleted.

Categories of personal information (CCPA / CPRA disclosure)

The table below summarizes the statutory categories of personal information we collect, why, and to whom we disclose it. We have not sold or "shared" (as those terms are defined under the CCPA/CPRA) any personal information, and we do not use personal information for cross-context behavioral advertising.

Sensitive personal information (CPRA). We collect health information when the Primary User or family use the optional health and medicine features, and photos, emails, voice, and conversations can contain sensitive personal information (for example, the contents of mail, a medication label, or a financial document). We use sensitive personal information only to provide the service the Primary User and family asked for — for example, explaining what was photographed, medicine reminders, and letting family know when the Primary User may need them (§8F) — to keep people safe (§8A), and to secure the service. We do not use it for advertising, and we do not sell or share it. You may contact us with any request about it.

California residents (CCPA / CPRA)

• Right to know / access the categories and specific pieces of personal information we collect, the sources, the purposes, and the categories of third parties to whom we disclose it (summarized in the table above)
• Right to delete personal information we hold about you
• Right to correct inaccurate personal information
• Right to opt out of the sale or sharing of personal information — we do not sell or share personal information, so there is nothing to opt out of. You can still record an opt-out on Your privacy rights, and a Global Privacy Control signal is recorded as one (see "Other US states")
• Right to limit the use of sensitive personal information — as explained above, we use it only for the purposes described; you may contact us with any request
• Right to non-discrimination for exercising any of these rights. We do not offer financial incentives in exchange for personal information.
• You may designate an authorized agent to act for you

New York residents

New York does not currently have a comprehensive consumer-privacy-rights law, but we honor the same rights framework available to residents of states that do: you may request access, deletion, and correction of your data by emailing support@mimicare.ai. We will respond within 30 days.

We additionally comply with:

• NY SHIELD Act (NY General Business Law §899-bb) — reasonable administrative, technical, and physical safeguards to protect the confidentiality, integrity, and availability of your private information.
• NY GBL §899-aa (data-breach notification) — see §11 below.
• NY Stop Hacks and Improve Electronic Data Security Act — incident-response practices documented in our Written Information Security Program.
• NY General Business Law Article 48 and the Attorney General's rules (13 NYCRR Part 700) — from January 1, 2027, the optional one-time age check described in §2 (two ways to do it, a review of an under-18 or unclear result, and a five-year record), and a few conversation features kept off until it is passed.

Other US states

Residents of Virginia, Colorado, Connecticut, Utah, Texas, Oregon, Montana, Iowa, Delaware, New Hampshire, New Jersey, Indiana, Tennessee, Kentucky, Maryland, Minnesota, and Rhode Island have substantially similar rights under their state privacy laws (to access, correct, delete, obtain a portable copy of, and opt out of the sale or targeted-advertising use of personal information). You may submit requests through the family dashboard or by emailing support@mimicare.ai. We honor Global Privacy Control (GPC) signals: when you are signed in to app.mimicare.ai and your browser sends a GPC signal while you use Your privacy rights, we record it on your account as an opt-out of sale and sharing — though we do not sell or share for cross-context behavioral advertising.

Right to appeal. If we decline your privacy request, you may appeal by replying to our decision or emailing support@mimicare.ai with the subject line "Privacy Appeal." We will respond to your appeal in writing, with our reasons, within the time your state's law allows (generally 45–60 days). If we deny your appeal, where your state provides one we will give you a method to contact your state Attorney General to submit a complaint.

Deceased users

If the Primary User has died, the Account Holder (or an authorized estate representative) may:

• Report the death by emailing support@mimicare.ai; our team records it. This closes the Primary User's seat and signs out its devices. If no other seat on the account is still in use or being set up, it also stops the subscription from renewing, with no refund: a subscription paid by card at app.mimicare.ai or through Google Play — including one that is paused or on hold — is canceled automatically at the end of the period already paid for, and stays available until then. Apple lets only the person who bought an App Store subscription cancel it, so we email the Account Holder the steps to cancel it in their device's Settings. If another seat on the account is still in use or being set up, the subscription continues unchanged; if a seat that was still being set up is never finished, we stop renewal when its invitation expires and email the Account Holder. If a seat on the account is in use again later — the report is withdrawn, or a seat is added or reopened — we turn card renewal back on if the paid period has not ended (unless a plan change the Account Holder had scheduled cannot be scheduled again: then renewal stays off and we ask them to choose their plan again), and email the Account Holder what that means for their plan, including when a card subscription renews and how to cancel it, and how to restore a Google Play or App Store subscription. (See the Terms of Service, §5.)
• Request a copy of the Primary User's data through the rights process above.
• 60 days after the death is reported, the seat's personal data — helps, photos, recordings, conversations and memories, health and medicine information, reminders, and emails — is deleted, except records we must keep for legal, billing, or safety reasons (including the record of AI conversations, §8J), and anything under a legal hold. The Primary User's first name stays as the closed seat's label in the family's list.

A new Account Holder may assume control of an account with documentation (power of attorney, executor letters, or guardianship order) emailed to support@mimicare.ai.

Biometric information (Illinois, Texas, Washington)

We do not collect, create, or retain biometric identifiers or biometric information as defined under Illinois BIPA, Texas CUBI, or Washington HB 1493:

• We do not extract, store, or compare voiceprints.
• Our AI describes document photos only — it does not analyze or describe facial features, hair color, eye color, ethnicity, or any other identifying personal characteristics of persons who incidentally appear in submitted photos.
• We do not perform facial recognition.

If biometric identifiers ever incidentally appear in a submitted photo, we do not extract or store them.

The one-time age check (§2). Only in a place whose law asks for an age check — today only New York — and only after the Primary User agrees on the check's screen, our provider Didit analyzes a picture of their face to estimate their age and check that it is a live person. If the Primary User chooses a photo ID instead, Didit also compares their face with the photo on the ID, only to confirm the ID is theirs. The pictures are used only to check age, never to identify the Primary User or to search for anyone. Didit deletes them right after the result, and we never receive the pictures or any measurement of the face — only the result.

10. Children

Mimi Care is designed for adults (a trusted contact setting up for another adult who will be the Primary User). The service is not directed to children, and we do not knowingly collect personal information from children under 13 (or, where applicable state law sets a higher age for certain protections, under that age). We do not knowingly sell or share the personal information of consumers under 16. If you believe a child's data is in our system, email support@mimicare.ai and we will delete it promptly.

Mimi Care is for adults 18 and over. Where a law requires it, Mimi Care offers the one-time age check described in §2; if a check shows someone is under 18, a few conversation features stay off on their phone, the Account Holder is emailed, and they can ask us to review the result.

11. Security

• TLS 1.3 in transit
• Application-layer envelope encryption (AES-GCM-256) for helps, conversation recordings, Companion notes and memories, daily activity, medicine details, reminder and calendar text, and the subject and body of email delivered to the Primary User's Mimi Care address — under one platform key for consumer accounts, and a separate key for each business (B2B) customer. The records listed in §3 as not encrypted rely on the security of our database and file storage.
• Secrets stored in Cloudflare Worker secrets and Doppler
• The record of AI conversations (§8J) is encrypted to a key kept offline, which our servers do not have
• Our AI request logs (§5) hold no content sent to or from the AI; access to them requires access to Mimi Care's Cloudflare account
• Rate limiting, abuse detection
• Principle of least privilege on admin access; staff support sessions and access to protected information are audit-logged (§6)
• Security incidents trigger the response plan in our Written Information Security Program

No system is perfectly secure. Mimi Care is a US-only service. If we learn of a breach affecting your data:

• NY SHIELD Act (NY General Business Law §899-aa): we notify affected NY residents, the NY Attorney General, the NY Department of State, and the NY Division of State Police without unreasonable delay after discovery, consistent with §899-aa(2).
• Other US state laws: we follow each state's specific timeline (most require "without unreasonable delay"; some specify 30–90 days).
• HIPAA: we are not a HIPAA covered entity on the consumer product. Where we operate B2B workloads for HIPAA-covered entities under a Business Associate Agreement, we follow HIPAA Breach Notification Rule requirements (45 CFR §§164.400–414).

12. Legal framing

Mimi Care is an informational tool. It is not medical advice, legal advice, financial advice, or tax advice. See our Terms of Service and AI Disclosure for more.

On the consumer product we are not a HIPAA covered entity or Business Associate. For B2B tenant deployments (home-health, hospice, PACE, payers, hospitals), we may act as a Business Associate under a signed Business Associate Agreement with the tenant covered entity.

13. Changes

We will post changes to this policy on mimicare.ai/privacy and update the "Last updated" date. Material changes will be announced in-app at least 30 days before they take effect and may require re-acceptance per Section 13A of the Terms of Service. A version that only corrects our description of how Mimi Care works, or that can only lower what you pay or add a protection — never raise a charge or take away a right — takes effect when it is posted. Compared with version 2.8, version 2.9 corrects how a death is reported (by email to support, not in the app), and can only lower what you pay or add protections: a reported death or an account deletion can now stop the subscription from renewing — and if a stop after a death report is undone, a card subscription returns to the renewal the Account Holder set up, on the same plan and price (or, if a plan change they had scheduled cannot be scheduled again, with its renewal left off), and we email them when that happens — and reminder and calendar text is now stored encrypted. It takes effect when it is posted.

14. Contact

Questions, complaints, or requests: support@mimicare.ai

BNC Solutions LLC 418 Broadway, Ste. N Albany, NY 12207

A Spanish-language informational translation of this Policy is available at mimicare.ai/privacy?lang=es. The English version is the legally binding document.