EaseKid Dict is a picture dictionary for children. A child can type the letters of a word they have met on a page, or say a word out loud, and get a picture, the word read aloud, and a meaning written for their age.
Every part of it is free. There is no subscription and no in-app purchase, so nothing here depends on what anyone has paid for.
EaseKid Dict does not collect names, email addresses, or any information that identifies you or your child. No accounts are required.
When a child taps "Say it", Apple's speech recognition turns what they say into text. The app asks for this to happen on the device itself, and on devices where iOS supports that, it does and the audio goes nowhere. Where iOS does not support it, Apple's speech service handles the audio under Apple's privacy policy. We never receive the audio, and nothing records it.
This is the same for both ways in: the letters the child typed, or the text Apple's recognition heard — never the recording. On "Say it" that text is usually one word, but it is whatever the child said: a child who describes a word instead ("the yellow thing in the sky") sends that phrase, and Apple's recognition can return up to three guesses at it.
The typed word, or the word the child picks, is sent to our server, which answers with the dictionary's entry for it. The device keeps only a word list (below), so it answers by itself only when the server cannot be reached. When the dictionary has no entry, or the child described a word rather than saying it, our server asks a large language model for the words it might be and a simple meaning. We use DeepSeek, with Google Gemini as a fallback when it is unavailable. This happens for everyone; there is no paid tier and nothing is held back.
The request carries that text, the reading level set in settings, and a security token (below), and nothing else: no account, no device identifier, no name, and nothing that says who asked. The reading level is one of three bands – K-1, 2-3 or 4-5 – never the grade itself. Our server uses it only to choose how simply the word is explained, and does not save it, count it or log it. The app cannot tell what a child chose to say, so if a child says their own name or something personal into "Say it" and the device has no word for it, that text is sent the same way. The spoken text itself is not saved; only the word worked out from it is, as described below.
The word, picture and meaning may then be saved on our server so the next child who asks for the same word gets an instant answer. What is saved is the dictionary entry itself — never the recording, never who asked, and never anything that could identify a child.
So that only EaseKid Dict, and not some other program, can use our server, the app proves it is the genuine app using Apple's App Attest. Once per install, Apple confirms to our server that the request comes from an unmodified copy of EaseKid Dict on a real Apple device, and our server gives the app a token that lasts one day.
The token is random. It contains nothing about you, your child or the device, and it is used only to keep the server safe from misuse: to tell genuine requests from others, and to stop one install from sending far more requests than a child could. We do not log it, we do not link it to the words looked up, and it stops working after a day. Apple's part is covered by Apple's privacy policy.
The app keeps a list of words, each with a picture and a short meaning, so it can suggest words as a child types and still help when there is no network. The list is stored encrypted on the device. Each time the app opens it asks our server whether there is a newer list, and downloads it if there is.
This request contains no information about you or your child – nothing derived from anything that was said or typed. It carries the security token and a tag naming the version of the list the app already has, which is the same for every install holding that list. If the check fails, the app keeps using the list it has.
Some words have a drawn picture, kept at img-dict.easekid.com. Once a day the app downloads the list of pictures there, the same list for every install. When it shows a word that has a picture, it downloads that picture and keeps a copy for a while, so it is not downloaded every time.
These requests carry no information about you or your child – no security token, no account and no device identifier. A picture request does show which word's picture is wanted, just as a lookup shows which word was asked for, and like a lookup it is not linked to anyone.
We do not use advertising networks or sell any data to third parties.
EaseKid Dict is designed for use by young children under parental supervision. We do not knowingly collect personal information from children under 13. No user accounts, profiles, or persistent identifiers are created.
We do not store personal data. Words looked up are saved on our server as ordinary dictionary entries — a word, a picture and a simple meaning — so that later requests are instant and so that reviewed words can be added to everyone's word list. These entries contain no identifying information and are not linked to whoever asked.
Each entry also carries a count of how many times that word has been asked for, which is how we decide which words to write next. It is one number against a word, with no time, no session and nothing about the person who asked — a count of a word, not a record of a lookup.
Questions about this policy? Email us at [email protected].