PRIVACY FIRST · V2.2
測試版隱私與資料處理說明
版本 2.2|2026-11-18
重要:這是開發測試版說明,不是已完成法律審核的正式隱私政策。公開營運前,仍須明確列出實際資料控制者、聯絡窗口、適用地區、保存期限與當事人權利流程。
一、裝置端功能
訊息分析、SafePay、ScamGraph、TrustReceipt 與情境圖鑑在使用者的瀏覽器內運行,使用規則比對與固定情境線索;本版不接 OpenAI、Gemini、Cloudflare Workers AI 或其他付費生成式 AI API。這不是經真實案件訓練、可保證準確率的 AI 模型。
App 不主動讀取通訊錄、通話紀錄、簡訊或麥克風;不含廣告追蹤 SDK。SafePay 與 ScamGraph 在檢查後清空原始訊息;單獨的訊息分析輸入框則保留文字直到你手動清除、重整或離開。任何原始訊息均不包含於安全報告;本次頁面時間線、報告及回報草稿預設只存於執行記憶體。下載的檔案可能留在個人裝置,須由使用者自行管理。
二、需另外同意的非 AI 網路功能
單筆電話查詢須使用者明確勾選後才會透過 HTTPS 傳送號碼至已完成部署的伺服器。正式資料庫尚待驗證,不能將查無結果解讀為安全。與伺服器互動的網路基礎設施可能產生必要連線紀錄;不可承諾完全不留任何網路痕跡。
可疑號碼私人回報 API 預設關閉。如依法完成營運設定、審核程序及人機驗證後開放,才可能儲存單筆號碼與分類以便人工審核;不會因檢舉直接公開、宣稱犯罪或自動封鎖。測試資料架構設計 30 日到期清理,但仍需正式流程驗收。
三、資料控制與安全限制
使用者可在 App 的隱私管理清除本次頁面輸入與草稿。若未來送出至伺服器,則須另依正式公告之權利行使流程申請查閱、更正、刪除或停止利用。安全分析可能誤判或漏判,也不能取代銀行與政府官方查證。
四、上線前待完成
正式營運主體及聯絡方式、法律告知與同意、資料最小化、申訴更正機制、權限審核、網站與原生應用資安測試、商標檢索,以及第三方正式法遵審查。
緊急防詐
台灣使用者如遇疑似詐騙,請主動查證 165 官方網站 或聯絡銀行官方客服。SecureRing 並非警政機關。
PRIVACY FIRST · V2.2
Preview privacy and data-handling notice
Version 2.2 | November 18, 2026
Important: This is a development-preview notice, not a legally reviewed production privacy policy. Public operation requires the actual operator's identity, contact details, regions served, retention periods and procedures for rights requests.
1. On-device functionality
Message analysis, SafePay, ScamGraph, TrustReceipt and pattern practice run inside your browser using deterministic rules and bounded scenario matching. This version does not connect to OpenAI, Gemini, Cloudflare Workers AI or any other paid generative AI API. It is not a real-case-trained or validated ML model.
The app does not access your contacts, call logs, SMS or microphone and contains no ad-tracking SDK. SafePay and ScamGraph clear source text after analysis; the standalone message scanner keeps text in its input field until you clear it, reload or leave. Original messages are never included in reports. Timelines, reports and draft call reports exist in page memory only. Downloaded files remain under your own device's control.
2. Separately consented non-AI network features
Single-number lookup transmits one voluntarily entered number to a configured server over HTTPS only after explicit consent. A properly reviewed directory is not yet deployed. No result should be interpreted as proof of safety. Infrastructure providers may process essential connection metadata; we cannot promise zero network traces.
The private report API is disabled by default. It may be activated only after operator readiness, review procedures, privacy notice, consent and human-verification checks. Reports must not automatically appear publicly, establish criminal accusations or result in auto-blocking. A 30-day expiry is designed into the draft schema but still requires production validation.
3. Controls and limitations
Use Privacy & Settings to clear current-page inputs and drafts. Data already sent to a future configured server requires a separate formal rights process for access, correction or deletion. The safety engine can miss scams or generate false alarms and cannot replace independent bank or government verification.
4. Production prerequisites
Legal operator and contact channel, appropriate notices/consent, data minimization, correction and appeal workflows, access reviews, web/native penetration testing, trademark clearance and independent legal compliance review.
Urgent help
In Taiwan, check the official 165 website or contact your bank using official information. SecureRing is not a government or police service.