befall Privacy Rights and AI Training Opt-Out Guide
This guide explains access, correction, deletion, portability, objection, app-scoped training opt-out, verification, authorized agents, appeals, and response timing.
Access and information
You may ask whether the operator processes personal information about you and request access to information covered by applicable law, including relevant categories, purposes, source types, recipient types, and available copies. A response may omit another person's information, security details, privileged material, or other content that law permits or requires us to withhold.
Correction
You may request correction of inaccurate personal information, including eligible account or age details, subject to verification and record-integrity requirements. We may retain a documented historical value when law, fraud prevention, transaction integrity, or a dispute requires it, while marking the current verified value appropriately.
Deletion
You may request deletion of covered personal information. We may retain only what applicable law permits or requires for security, fraud prevention, transaction records, legal obligations, claims, or a valid preservation hold. Account deletion, individual conversation deletion, a privacy deletion request, and training opt-out are separate choices and should be identified clearly.
Restriction, objection, and state choices
Where applicable, you may object to or request restriction of certain processing, limit use of covered sensitive information, and opt out of qualifying sale, share, or targeted advertising activity. We will not unlawfully discriminate because you exercise a state privacy right. Vendor verification must finish before the service publishes a definitive classification of each SDK disclosure.
Portability
Where applicable law requires it and the information is technically portable, you may request a copy in a commonly used format. Portability does not require disclosure of another person's data, protected security signals, proprietary model parameters, or information that cannot lawfully be provided.
AI training opt-out
You may opt befall out of future model-improvement use without closing the account. After the opt-out becomes effective, the user's data must be removed from pending training queues, uncompleted labeling jobs, and separable datasets, and later-created data cannot enter those workflows. Any copy retained by law or for security is quarantined and not used for training. The operator cannot promise to remove a contribution already absorbed into a completed model, but must act when law or separable lineage requires it.
Identity verification
We verify only what is reasonably necessary for the request, considering its sensitivity and risk. Verification may use existing account access or matching information; it should not demand a new identity document when a less intrusive method is sufficient. Information collected for verification is restricted to that purpose and related legal records.
Authorized agents and appeals
An authorized agent may submit a request where applicable law permits, subject to reasonable proof of authority and, when allowed, direct confirmation from the user. If we deny or limit a request, we explain the basis and provide an appeal method. A denied appeal may also be directed to the state regulator or other authority identified by applicable law.
Response timing and contact gate
We acknowledge, investigate, and respond within the period required by applicable law, using any lawful extension with notice and an explanation. The final request mailbox and web path must be verified before release, so this draft states no invented address or fixed deadline that would conflict with a controlling state rule.
Document history
- 2026-07-27-draft.1 · 2026-07-27 — Initial PROJECT82 regional legal draft.
Last updated 2026-07-27 · content d2e0c6e99af4 · 대한민국 · 한국어 · 日本 · 日本語