Privacy Policy

1. Who we are

gugudan ("gugudan", the "app", "we", "us") is a voice-based multiplication-tables practice app for elementary-school children. The app is operated by Semolab (세모연구소), a business registered in the Republic of Korea. You can reach us at hello@gugudan.app.

This policy explains what information the app handles, with a focus on children, and how parents and guardians can exercise their rights. It is written to follow the U.S. Children's Online Privacy Protection Act (COPPA).

2. Our approach to children's data

gugudan is designed for children, so we collect as little as possible:

• There is no sign-up and no account. We do not collect a child's name, age, birthday, email address, phone number, contacts, location, or photos.
• The app works fully in keyboard mode, in which nothing leaves the device.
• The only information that ever leaves the device is the child's spoken answer, and only when a parent has enabled the optional voice-grading feature (see Sections 4 and 5).

3. Information we handle

3.1 Stored only on the child's device

The following are stored only on the device and never sent to a company server. They contain no name, age, or other identifying details, and are removed when the app is uninstalled:

• Learning records (problems solved, correct/incorrect history)
• In-progress session state and per-table study statistics
• Language setting
• A consent record (the date/time of consent, the items consented to, and the disclosure version)

3.2 Handled only when voice grading is ON

• The child's spoken answer (audio). A short spoken answer to a multiplication question (for example, "twenty-one"). It is used only to convert speech to text so the app can grade the answer.
• A device-integrity token (Firebase App Check). A token used only to block automated abuse of our relay. It is not personally identifying and is generated only while voice grading is enabled.

We do not create a voiceprint, perform speaker identification or any biometric analysis, use the audio to identify the child, profile the child, or use the audio to train any model.

4. How the spoken answer is processed

When voice grading is on, the child's spoken answer is sent over an encrypted (TLS) connection to gugudan's own relay server (operated by Semolab on Google Cloud, in the United States) and then to Google Cloud Speech-to-Text (United States) for transcription only.

• The audio is processed ephemerally — held in memory only for as long as it takes to transcribe the answer in real time.
• It is deleted immediately after transcription. Our relay is stateless and stores no audio on disk, in logs, or in any database, and the transcription service is configured not to retain, log, or train on the audio.
• There is no behavioral advertising, profiling, or tracking at any point.

5. The optional voice feature and parental consent

Voice grading is off by default and is a setting a parent turns on. When it is on, the only information that leaves the device is the child's short spoken answer, used solely to convert speech to text so the app can grade it (Sections 3.2 and 4).

Under COPPA, a child's audio that is collected solely as a replacement for the written answer, used only for that purpose, not used to identify the child, and deleted immediately after it is converted to text may be handled without separate verifiable parental consent — the COPPA "audio file" exception. gugudan is built to meet these conditions: the spoken answer stands in for the written one, is used only for transcription, is never used to create a voiceprint or to identify or profile the child, and is deleted immediately after transcription (Sections 3.2, 4, and 6).

A parent stays in control at all times:

• The feature is off until a parent turns it on, and the app is fully usable in keyboard mode without it.
• A parent can withdraw at any time by turning voice grading off in the app's Settings, which stops any further transfer of the spoken answer immediately.

If we ever change the voice feature so that this exception no longer applies, we will obtain verifiable parental consent before that change takes effect.

6. Retention and deletion

Our data-retention policy is to keep as little as possible, for as short a time as possible:

• Spoken answer (audio): deleted immediately after transcription. We retain none of it.
• On-device data (learning records, statistics, settings, consent record): stored only on the device and removed when the app is uninstalled. We do not store this on a server.
• Consent record: the device-local record is kept until the app is uninstalled. If a de-identified server-side receipt of consent is kept, it is retained for no more than one year and then deleted. Such a receipt contains only the date/time of consent, the items consented to and the disclosure version, and the fact that the voice feature was enabled — never a phone number, name, or government identifier.

We store no personal information about a child on a company server, so there is no server-side profile to retrieve or purge.

7. Parental rights

As a parent or guardian, you may:

• Review the information the app handles about your child (the categories are listed in Section 3);
• Refuse further collection by turning off voice grading in Settings, or by uninstalling the app;
• Delete on-device data by uninstalling the app. The spoken answer is never retained, so there is nothing to delete after transcription.

Because we keep no personal information about your child on a server, remote retrieval is limited by design. For any request or question, contact us at hello@gugudan.app and we will respond without undue delay.

8. Third parties and processors

We do not sell children's personal information and do not disclose it to third parties for their own purposes. The spoken answer is shared only with the service providers below, solely to provide the voice-grading feature:

9. Security

• Data in transit is encrypted with TLS.
• The relay is stateless and stores no audio.
• Learning records are stored on the device only.
• Device-integrity checks (App Check) help block abuse.
• We minimize what we collect — no account, no identifying details.

10. No tracking, ads, or analytics

The app uses no cookies, no advertising, and no analytics or crash-reporting tools. This website also uses no analytics or tracking. The only automatically generated value is the device-integrity token described above, and only while voice grading is on.

11. Changes to this policy

We may update this policy as the law, our service, or our legal review changes. When we do, we will post the updated policy here with a new effective date. For any change that materially affects how we handle children's information, we will obtain parental consent where the law requires it.

12. Contact

Semolab (세모연구소) · hello@gugudan.app