Terms on Collection of Host Voice/Content Data
and AI/Marketing Use

Article 1 (Purpose and AI Transparency Notice)

These Terms define the rights and obligations arising in the course of Labuchat Inc. (the "Company") collecting, using, training on, and using for marketing the Host's voice data and the content created by the Host in connection with the Service.

Article 2 (Definitions)

1. Voice Data means audio information generated, transmitted, and stored in the course of the Host using the Service, and digital information such as feature points extracted therefrom.
2. Content means all voice, video, image, and text information created or registered by the Host within the Service, including calls (recordings), live broadcasts (recordings), feeds, messages, and images registered by the Host to the Service such as profile photos and character images.
3. Guest means a user who interacts with the Host, such as in a 1:1 call.
4. Pseudonymization means processing that deletes or replaces part of personal information so that a specific individual cannot be identified without additional information.
5. Derivative Works means advertisements (such as short-form videos), AI training models, algorithms, and the weights of such models created by the Company on the basis of voice data and content.

Article 3 (Mandatory Consent: Service Provision and Safety) REQUIRED

The Host acknowledges and agrees that the following are essential for the performance of the core functions of the Service and for its safe operation. Processing under this Article is limited to the minimum necessary for service provision such as call connection and settlement.

1. Service Operation and Security: Real-time call quality monitoring, noise removal, detection of fraudulent use (including identity verification), and verification of compliance with community guidelines.
2. Technical Optimization: Data processing for stable connection and basic quality improvement of the Service, such as improving speech recognition (STT) accuracy.

Article 3-2 (Optional Consent: Content Marketing Use) OPTIONAL

Where the Host consents, the Company may edit and process the content created by the Host (calls, live streams, feeds, etc.), images registered by the Host to the Service such as profile photos and character images, and the Host's own voice and video for service promotion purposes, and use and distribute them in short-form videos (Shorts), social media advertising, external banners, and service branding campaigns. For this, the Host grants the non-exclusive license under Article 8.

Article 4 (Guest Protection and Third-Party Information Handling)

1. Where the Company uses for marketing content in which the Host and a Guest both participated, such as a 1:1 call, the Guest's voice, video, and personally identifiable information will be used only after being processed so that the specific individual cannot be identified through technical measures such as pseudonymization, voice alteration, synthetic voice replacement, and personal-information masking.
2. The Host acknowledges that consent under these Terms is effective only with respect to the Host's own voice, likeness, and content and does not affect the rights of the Guest.
3. The Host shall take care not to expose, within the content, the personal information of a Guest or a third party (real name, contact, address, etc.) in a form unsuitable for marketing use.

Article 5 (Optional Consent: AI Clone Training and Retention — Sensitive Information) OPTIONAL

Where the Host consents, the Company may use the Host's voice data to train and create an AI voice/persona ("AI Clone") model that reproduces the Host's voice, and may retain and operate such model. Because the Host's voice feature information is treated as sensitive information equivalent to biometric data, this consent is obtained separately from other consents. The Host may refuse consent to this Article, and even upon refusal may still use the basic Service as a Host.

1. Purpose: Training/creation of the Host-voice-based AI Clone and retention/operation of that model (providing service features such as voice synthesis (TTS) and persona-based NLU). This purpose is independent of the use/monetization consent under Article 5-2 and is maintained regardless of whether monetization consent is given.
2. Items: The Host's call voice and the voice feature information extracted therefrom, and the trained model weights.
3. Retention Period: Retained for the period the Host qualification is maintained; upon withdrawal of this consent, account withdrawal, or a deletion request, handled in accordance with Article 9.

Article 5-2 (Optional Consent: AI Clone Use and Monetization) OPTIONAL

Where the Host consents to this Article, the Company may use the Host's AI Clone trained/created under Article 5 for calls, content, marketing, and the like, and generate revenue therefrom. This consent is a separate optional consent distinct from Article 5 (training and retention).

1. Revenue generated by AI (including AI Clones) belongs to the Company. The Company may, pursuant to its policy, return to the Host a portion of the revenue generated by that Host's AI Clone; the rate, whether to pay, and any changes are determined by the Company (a discretionary benefit, not a guaranteed right).
2. The Host may withdraw this consent at any time. Upon withdrawal, the Company ceases the new creation, distribution, and monetization use of that Host's AI Clone thereafter (prospective effect). However, as long as the Article 5 (training and retention) consent is maintained, the trained model is retained, and if the Host consents to this Article again, use may resume from that point.
3. AI that the Company creates independently of any Host is not subject to this Article. Recovery of AI works already distributed/posted externally is governed by Article 9.

Article 6 (Retroactive Application and Effect of Optional Consents)

1. Where the Host consents to Article 3-2 (Content Marketing Use) or Article 5 (AI Clone Training and Retention), such consent applies not only to content created after the time of consent but also retroactively to the Host's existing content created and stored before the time of consent (call recordings, live recordings, feeds, messages, profile photos and character images, etc.).
2. Use of existing content shall equally comply with the Guest protection measures under Article 4.
3. The consents under Articles 3-2, 5, and 5-2 are all optional; even if the Host does not consent or withdraws consent, there is no restriction on basic Host activity such as calls and settlement. Upon withdrawal, the retroactive use under the relevant consent also ceases prospectively.

Article 7 (AI Generation Notice and Right to Explanation)

1. The Company implements labeling (such as watermarks) by which virtual voice outputs generated through AI technology can be recognized as AI-generated, as provided by applicable laws.
2. The Host may request an explanation of the general principles and criteria of the AI training in which the Host's data is used. The Company will respond in good faith to the extent it does not infringe trade secrets.

Article 8 (Attribution of Rights and License)

1. The Host holds the moral rights to the Host's own voice data and content, and the Company is granted a non-exclusive license to use such data (reproduce, modify, create derivative works, distribute, etc.) within the scope of the purpose of these Terms. Such license is deemed included in the economic benefits the Host obtains within the Service and in the consideration for providing the Service.
2. The intellectual property rights to the AI training models, algorithms, and all derivative works developed by the Company on the basis of voice data and content belong to the Company.

Article 9 (Data Protection and Destruction)

1. Upon the Host's withdrawal of the Article 5 (AI Clone Training and Retention) consent, account withdrawal, or a request to delete or suspend processing of personal information, the Company, in accordance with personal information protection laws, destroys without delay (and no later than within 30 days) the relevant training data and identifiable feature information (speaker embeddings, identification keys, identifier mappings, etc.) so that they cannot be restored or reproduced. Where only the Article 5-2 (Use and Monetization) consent is withdrawn, monetization use ceases, but the training data is not subject to destruction so long as the Article 5 consent is maintained.
2. Original call voice is, in principle, destroyed without delay once the AI Clone (feature information) has been created, as its collection/use purpose is deemed achieved. However, where there is a separate legitimate basis such as dispute response, fraud verification, or a statutory retention obligation, it may be stored encrypted and separately from other data for the period permitted by applicable laws to the extent necessary for that purpose, and is destroyed without delay once that purpose ends.
3. The following are confirmed to be excluded from destruction because identifiability as personal information has been lost or physical recovery is impossible.
   • a. Model weights and parameter information reflected in AI model training in a state where the original data of a specific individual cannot be technically restored. However, speaker embeddings, identification keys, and the like that can identify or reproduce a specific individual do not fall under this exception and are destroyed in accordance with Paragraph 1.
   • b. Marketing works whose creation and external distribution have been completed and which are beyond the Company's reasonable control and cannot be recovered. The Company will, however, take deletion measures within the scope it can control.
4. The destruction obligations under this Article and the Host's rights to withdraw consent and to request deletion or suspension of processing are not limited by these Terms or any individual consent, and the procedures for exercising them shall not be more difficult than at the time of consent collection.

Article 10 (Security Measures)

1. When pseudonymizing voice information, which is unstructured data, the Company complies with the latest guidelines and separately stores or technically alters information with high identifiability, such as voice prints.
2. When using voice data and content for AI training or marketing, the Company physically and logically separates and manages the original data and the processed data, and implements technical and managerial protection measures such as encrypted storage and access-rights restrictions.

Article 11 (User Age Restriction)

This Service may be used only by users aged 14 or older. Children under the age of 14 cannot register as Hosts, and the Company immediately destroys the personal information of any user confirmed to be under the age of 14.

Article 12 (Data Storage Location)

For service stability and compliance with each country's personal information protection laws, the Company stores users' data on servers located in the service region.

• Korea users: AWS Seoul Region (ap-northeast-2)
• US users: AWS Oregon Region (us-west-2)
• Taiwan users: AWS Hong Kong Region (ap-east-1). As no AWS region is available within Taiwan, data is stored in the geographically closest Hong Kong region, which constitutes entrusted processing for service provision.

Addendum

These Terms take effect on July 3, 2026. This amendment clarifies that content subject to marketing use includes images registered by the Host to the Service, such as profile photos and character images (no substantive change to rights or obligations). Under the previous amendment (June 23, 2026), Content Marketing Use was changed from mandatory to an optional consent, and AI Clone consent was split into Training and Retention (Article 5) and Use and Monetization (Article 5-2).