Effective November 5, 2026.
This policy describes how Kinlight handles personal information for both of our products: Kinlight, an AI companion for families, and Kinlight for Schools, used by students, teachers, and school staff at partner schools. Where something works differently for schools, we say so.
Kinlight is operated by KinlightAI, Inc. ("Kinlight," "we," "us"), which took over the service from Kinlight LLC. You can reach us at hello@kinlight.ai, by phone at (385) 350-3956, or by mail at KinlightAI, Inc., 972 W Sedge Ct., Farmington, UT 84025.
Kinlight is used by children under 13, and we built it around that from the start.
Family accounts. A Kinlight account can only be created by a parent or legal guardian. A child never signs up on their own, and a child is never asked to type their own age or birthdate — a parent sets up each family member's profile themselves and chooses their age band (younger child, teen, or adult). Before a parent can use the family dashboard, they must check a box confirming they are the child's parent or legal guardian, type their full legal name, and agree to the terms shown to them — we keep a record of that agreement.
School accounts. A school's administrator — not Kinlight — sets up the school's account, rosters students into classes, and links each student to their parent's email address. We then send that parent a private, secure, one-time link (not the school's own login system) asking them to review and sign consent for their child's use of Kinlight at school, specific to that child. We keep a record of who signed, and when.
We do not knowingly collect personal information directly from a child beyond what a parent or school administrator has provided in setting up that child's profile.
Messages sent to Kinlight are processed to generate a response. Most ordinary conversations are never reviewed by anyone and are not shown to a parent, a teacher, or Kinlight staff. A short record of a child's most recent conversation is kept temporarily so they can pick up where they left off, and is deleted automatically after 30 days of inactivity. Recent conversations are also kept a little longer — roughly nine months — to support the safety review described in Section 5, after which they're automatically deleted.
If you use Kinlight's "read aloud" feature, your own device reads the text out loud itself — nothing is sent anywhere for this. Kinlight also has a separate, optional feature that can turn a response into spoken audio using a third-party text-to-speech service; this sends the text of that response (never your recorded voice) to that service to generate the audio clip. Kinlight does not record or collect anyone's own voice.
If you turn on notifications, we store a device token so we can deliver them to you — on an iPhone, through Apple's own push notification service; on Android, through Google's Firebase Cloud Messaging. Either way, the notification itself is a short, fixed alert and which screen to open, never a child's name, a category, or any message content, and we send it directly to Apple or Google, with no other company in between.
At a partner school, Kinlight for Schools also holds, as the school's own record: class rosters, staff roles, and the link between a student and their parent(s), all set up by the school's administrator; assignments, student submissions, grades (including a full history of any changes to a grade — never just the current value), and teacher feedback; attendance; and behavior records. These are kept as the school's records for as long as the school's account is active, and are not automatically deleted the way ordinary chat history is, since a school may need to keep them for longer. A behavior record's staff-written description is visible only to school staff; a parent sees the category and date, never the staff member's own notes.
Payments are processed entirely by Stripe or by Apple. Kinlight never sees or stores your card number.
Before any response is generated, every message passes through a safety check that doesn't depend on the AI itself — a fixed check for signs of an immediate crisis (for example, thoughts of self-harm) or a possible disclosure of abuse — alongside a second, AI-based check built to catch the same things when they're phrased less directly. A separate review then checks the AI's own proposed response before it's ever shown to a child. If a message shows signs of crisis, Kinlight responds with a consistent, pre-written message: it validates what the person is feeling, points directly to real help (in the US, 988 for the Suicide & Crisis Lifeline, any time, day or night, or 911 if someone is in immediate danger), and asks them to tell a parent, pastor, or another trusted adult. A possible disclosure of abuse is met with the same seriousness. See exactly how Kinlight's safety system works for the fuller picture. None of this depends on whether a parent or school happens to be watching at that moment — it runs the same way every time.
A parent on a family account can see, through their family dashboard, most messages flagged for a safety or values concern, along with a plain-language explanation of what was going on. Two categories are deliberately kept out of your dashboard and out of your child's visible conversation history altogether — a possible disclosure of abuse, and an image that may show physical abuse — so that a child's disclosure about something happening in the home is never shown to the person it may concern. If a message shows signs that someone may be in an immediate crisis, we always make this visible to you right away in your dashboard, and we may also notify you directly by email or an app notification. Ordinary, non-flagged conversations are never shown to a parent — not as a flagged item, and not as a readable transcript you can browse.
A teacher at a partner school can see grades, attendance, and general patterns for the students in their own classes — never another class's students, and never the content of a safety concern. Only a school's counselors and administrators, specifically configured by the school to receive this, see the actual content of a flagged conversation. A parent whose child attends a partner school sees their own child's grades, attendance, and behavior category (never a staff member's written description); a safety concern at school is handled directly by the school's own designated staff, consistent with the school's own policies, not shown through a parent's own Kinlight account.
A parent whose child attends a partner school can also get AI help drafting a message to their child's teacher. This is a drafting tool only — Kinlight has no way for a teacher to receive or read it; a parent copies and sends it themselves through a channel their school already uses.
A trusted adult named by a homeschooling family never sees any conversation, and has no ongoing access of any kind. They're contacted only if there's a genuine, immediate safety concern — the same message a parent would get — and they can decline this role at any time.
These companies process certain information on our behalf, only to perform the function listed, and are not permitted to use it for their own independent purposes:
We do not sell personal information to anyone, for any reason. Kinlight has no advertising of any kind, and we do not use conversation content, or any other information described in this policy, to serve ads.
We do not currently use conversation content to train or fine-tune any AI model, for anyone. If that ever changes, it will only be for people — or, for a child's profile, a parent choosing this on the child's behalf — who have been told about it in advance and have explicitly opted in; it will never happen by default, we will remove identifying details first, and for a school's students, it would additionally require the school's own agreement, separate from a parent's. We separately use basic, non-content usage data (described in Section 3) to improve Kinlight — that's a different, much narrower thing than AI training, and is handled the same way whether or not anyone has opted into anything above.
Beyond the specific periods already described above (30 days for a resumable conversation, roughly nine months for conversation logs supporting safety review, up to a year for a submitted assignment photo), a flagged safety record is generally kept for 90 days unless a parent chooses to keep it longer, or, for a school, a resolved safety matter is kept up to a year. School records — assignments, grades, attendance, and behavior records — are kept for as long as the school's account stays active, for the reasons described in Section 3.
A parent can delete their entire family account, or a single child's profile, directly from inside the app — this immediately and permanently deletes everything described above, including the technical safety log of messages and responses described in Section 3, with no waiting period and no way for us to recover it afterward. The only exception is a separate internal error log, used only when a safety check briefly fails to run, which may include the text of that one message; it isn't linked to any profile, and it is deleted automatically 270 days after each entry is written. A school administrator can do the same for a school's account, or remove a single student's account instead. If you'd rather not handle a deletion yourself, email us at hello@kinlight.ai and we'll take care of it for you — see our Support page for exactly what to include.
Sensitive content — flagged messages, saved conversations, and other personal information described above — is encrypted at rest. Access to a family's or a school's data is restricted to that family's or school's own authenticated account; no family or school can see another's information, and Kinlight staff do not review individual accounts except as necessary to maintain the safety and security of the service.
We may update this policy as Kinlight evolves. Material changes will be reflected with an updated effective date at the top of this page.
Questions about this policy, or a request to delete your data, can be sent to hello@kinlight.ai, by phone to (385) 350-3956, or by mail to KinlightAI, Inc., 972 W Sedge Ct., Farmington, UT 84025.