Privacy Policy
Last updated: June 29, 2026
This Privacy Policy describes how the Refreshing browser extension ("the Extension," "we," "us") collects, uses, stores, and shares user data. The Extension is designed for use on X (formerly Twitter) and provides AI-generated replies, batch unfollow, following/blocked-list export, keyword monitoring with auto-follow, and spam-account detection.
By installing and using the Extension, you consent to the data practices described in this Privacy Policy.
1. Data We Collect
1.1 Data You Provide
- NVIDIA API Key — You may optionally enter your own NVIDIA API key in the Extension settings to use AI features. This key is stored in
chrome.storage.sync and synced across your devices via your Google account.
- Persona List — You may configure a list of AI persona names used for reply generation. Stored in
chrome.storage.sync.
- Keyword Lists — You may configure keywords for the auto-follow monitoring feature. Stored in
chrome.storage.sync.
- Numeric Settings — Various numeric preferences (follow delay, monitoring interval, scroll count, rate limits, browse time ranges, follower thresholds, etc.). Stored in
chrome.storage.sync.
- UI Language Preference — Your selected extension UI language. Stored in
chrome.storage.sync.
1.2 Data Automatically Collected
- X Bearer Token — The Extension intercepts the Bearer token from X's own GraphQL API requests via the
webRequest API. This token is stored in memory and chrome.storage.session (cleared when the browser closes). It is used only to make first-party X API requests on your behalf (e.g., exporting your lists, blocking accounts). It is never transmitted to any third party.
- X CSRF Token (ct0) and User ID (twid) — Read from
document.cookie on x.com pages. Used only to authenticate X API requests within your own session. Stored in memory only; never written to persistent storage or transmitted externally.
- Device Fingerprint — A random 16-byte hex value generated per page load (not persistent). Sent only to the xhunt ranking service (see Section 6.4) when you opt in to the xhunt export feature.
1.3 Data Read from X (Twitter)
- Tweet Text, Usernames, and Images — When you use the AI reply feature, the Extension reads the tweet thread content (text, author usernames, and images as base64) from the page DOM. This data is sent to NVIDIA for AI processing (see Section 6.1) and is not stored persistently by the Extension.
- Following and Blocked Lists — When you export your following or blocked accounts, the Extension calls X's first-party GraphQL API to retrieve the list. Data includes screen names, display names, bios, follower counts, and account metadata. This data is processed in memory and saved as a CSV/JSON file locally by you. It is not uploaded to any Extension-operated server.
- Search Results (Keyword Monitoring) — When keyword monitoring is active, the Extension uses the Chrome DevTools Protocol (CDP) debugger to intercept X's SearchTimeline API responses. Extracted data includes screen names, display names, user IDs, bios, follower counts, verification status, tweet content, and relationship status. This data is stored in
chrome.storage.local for processing.
- User Profile Data (Spam Detection) — The spam-detection module reads user profile information (handle, user ID, bio, display name, avatar, follower/following counts, account creation date, relationship status) from X's GraphQL responses and page DOM. Detection results are cached in
chrome.storage.local.
- Timeline/Feed Data (Hot Posts) — The hot-posts feature intercepts X GraphQL responses to extract tweet metrics (views, likes, retweets, replies, bookmarks, full text) for local velocity calculation. This data is not transmitted externally.
2. How We Collect Data
- webRequest Header Interception — The
webRequest permission is used to capture the X Bearer token from outgoing X API request headers.
- fetch/XHR Hooking — In the page's main world, the Extension wraps
window.fetch and XMLHttpRequest to intercept X GraphQL API responses for spam detection and hot-posts features.
- Chrome DevTools Protocol (CDP) Debugger — The
debugger permission is used for (a) simulating mouse movements and clicks for batch unfollow and auto-follow actions, and (b) intercepting SearchTimeline network responses during keyword monitoring. When the debugger is attached, Chrome displays a "DevTools is debugging this browser" banner at the top of the page. The debugger does not read page DOM content directly; DOM scraping is performed by content scripts using standard DOM APIs.
- DOM Scraping — Content scripts read tweet text, usernames, images, and profile information from the page's DOM using
querySelector and innerText.
- Cookie Reading —
document.cookie is read to obtain the X CSRF token and user ID for authenticating X API requests.
- User Input — Settings you enter in the Extension popup (API key, persona list, keywords, numeric preferences).
3. How We Use Data
- AI Reply Generation — Tweet thread content (text, usernames, images) is sent to NVIDIA's AI API (via our Cloudflare Worker proxy) to generate suggested replies.
- Auto Follow-Back (Keyword Monitoring) — X search results are monitored for users matching your keyword and filter criteria. The Extension automatically follows matching users. When AI content filtering is enabled, candidate users' names, bios, and tweet text are sent to NVIDIA for evaluation.
- Batch Unfollow — The Extension automates the unfollow process by simulating mouse clicks on X's unfollow confirmation dialog via CDP.
- Following/Blocked List Export — The Extension retrieves your following/blocked lists from X's API and allows you to download them as CSV/JSON files.
- Spam Account Detection (MXGA) — The Extension analyzes X user profiles using local heuristic rules to identify potential spam accounts and can automatically block them. Only numeric X user IDs are sent to the public ranking service (see Section 6.5); no bios or tweet content are uploaded for this feature.
- Hot Posts Detection — The Extension monitors your feed for rapidly trending posts using local velocity calculations. No data is transmitted externally.
- Local Statistics Display — The Extension maintains local statistics (follow counts, block counts, monitoring stats) for display in the Extension UI. These are not transmitted to any server.
4. Data Storage
4.1 chrome.storage.sync (synced via your Google account across devices)
NVIDIA API key, persona list, model ID, preview count, keyword monitoring configuration (follow mode, AI filter mode, AI model, follower thresholds, scroll count, delays, rate limits, browse time, panel mode, keyword selections, keyword text), UI language preference, AI button position, risk acknowledgment flag.
4.2 chrome.storage.local (persistent on this device only)
Keyword monitoring state/stats/logs/accumulated data/rate queue/intercepted user data/session token, batch unfollow progress and pending configuration, export panel states, MXGA spam-detection data (blocked ID sets, blocklist records, per-user verdict cache, block queue, whitelist, skip status, stats), XVM settings and status.
4.3 chrome.storage.session (cleared when browser closes)
X Bearer token, keyword-monitor panel window ID, global X write-operation timestamps for rate limiting.
4.4 IndexedDB (service worker context)
Persona list database — stores persona names, version number, and last-check timestamp for the persona update mechanism.
4.5 localStorage (on x.com pages)
XVM panel position/minimize state, MXGA block-timing state, MXGA onboarding flag.
4.6 In-Memory Only (not persisted)
X CSRF token, X user ID, tweet text/images for AI processing, device fingerprint, intercepted GraphQL response bodies (after processing).
5. Data Retention
- chrome.storage.sync/local — Persists until you uninstall the Extension or manually clear Chrome storage/browsing data.
- chrome.storage.session — Cleared when the browser is closed.
- MXGA cache — Per-user verdict cache and blocklist are capped at 5,000 entries each; oldest entries are evicted automatically.
- Keyword monitoring logs — Capped at 500 entries; oldest entries are evicted automatically.
- In-memory data — Lost when the page is closed or the service worker restarts.
6. Data Sharing with Third Parties
The Extension shares certain data with the following third parties. We do not sell user data. All data sharing is strictly for functionality purposes.
6.1 NVIDIA AI API (integrate.api.nvidia.com)
| Trigger | Data Sent | Purpose |
|---|
| AI Reply (via Cloudflare Worker) | Tweet text, author usernames, tweet images (base64), AI persona name, your NVIDIA API key (or built-in key if none configured) | Generate AI-suggested replies |
| Keyword Monitor AI Filtering (direct connection) | Candidate user's display name, bio/description, tweet text, your NVIDIA API key | Evaluate whether a user matches content criteria before auto-following |
NVIDIA Privacy Policy: https://www.nvidia.com/en-us/about-nvidia/privacy-policy/
6.2 Cloudflare Worker (ai-server-worker.ai-refreshing.workers.dev)
Our self-operated Cloudflare Worker acts as a proxy between the Extension and NVIDIA's API. When you use the AI reply feature, the Worker receives the full request payload (tweet text, usernames, images, persona name, API key), reads the persona prompt from Cloudflare Workers KV, and forwards the request to NVIDIA. The Worker does not log or store your request data beyond transient processing.
Cloudflare Privacy Policy: https://www.cloudflare.com/privacypolicy/
6.3 DuckDuckGo (lite.duckduckgo.com)
When the AI reply agent determines that a web search is needed, the Extension sends a search query (generated by the AI model from the tweet context, not your direct input) to DuckDuckGo Lite. No personal information or X account data is included in the search request.
DuckDuckGo Privacy Policy: https://duckduckgo.com/privacy
6.4 xhunt KOL Ranking (kb.cryptohunt.ai)
When you export your following or blocked list and opt in to the "Include xhunt ranking" option, the Extension sends the list of X screen names (public usernames), the current page URL, and a random per-session device fingerprint to the xhunt ranking API to retrieve KOL ranking data.
| Data Sent | Purpose |
|---|
| X screen names (from your exported list), page URL, random device fingerprint, request signature | Retrieve KOL ranking scores for exported accounts |
6.5 MXGA Edge Service (x.zuoluo.tv)
The spam-detection module communicates with the MXGA Edge service for two purposes:
- Public Ranking Query — Sends numeric X user IDs (up to 100 per batch) to check public ranking data. No screen names, bios, or tweet content are sent.
- Whitelist Sync — Downloads the maintainer's whitelist of known-legitimate accounts every 6 hours. No user data is uploaded during this sync.
6.6 Cloudflare Pages (ai-server-page.pages.dev)
The Extension downloads the persona list and version information from Cloudflare Pages. This is a download-only operation; no user data is uploaded.
6.7 X (Twitter) First-Party API
The Extension makes requests to X's own API endpoints using your active session (Bearer token + CSRF cookie) for the following purposes: exporting your following/blocked lists, blocking accounts (spam detection), and triggering Bearer token re-capture. These requests are made as your own authenticated session and do not pass through any Extension-operated server.
7. How Third Parties Use Your Data
Each third party listed in Section 6 processes data according to their own privacy policies (linked in each subsection). We do not control how NVIDIA, Cloudflare, DuckDuckGo, xhunt, or the MXGA Edge service process data after receipt. We encourage you to review their privacy policies.
8. Data Security
- All data transmissions to external services use HTTPS/TLS encryption.
- Your NVIDIA API key is stored locally in
chrome.storage.sync and is only transmitted to NVIDIA (directly or via our Worker) when you actively use AI features.
- The Cloudflare Worker uses a built-in fallback API key pool when you have not configured your own key. This pool is shared among all users who have not configured a personal key.
- We do not sell, rent, or exchange user data for monetary gain.
- No analytics, crash-reporting, or advertising SDKs are included in the Extension.
9. Your Choices and Controls
- AI Features — You can choose not to configure an NVIDIA API key. AI reply and AI content filtering features will not function without a key.
- Keyword Monitoring — You can choose not to start keyword monitoring. You can disable AI content filtering (set mode to "Off").
- xhunt Export — The xhunt ranking feature is opt-in only. Do not check "Include xhunt ranking" if you do not want your screen name list sent to the xhunt service.
- Spam Detection (MXGA) — You can disable the spam-detection module in the Extension settings.
- Clearing Data — You can clear all Extension data by uninstalling the Extension or by navigating to Chrome Settings → Privacy and security → Clear browsing data → Cookies and other site data (this clears
chrome.storage). Keyword monitoring logs can be cleared from the Extension popup.
- chrome.storage.sync — Data synced via your Google account can be managed at chrome.google.com/sync.
10. Children's Privacy
The Extension is not directed at children under the age of 13. We do not knowingly collect personal information from children under 13. X's own Terms of Service require users to be at least 13 years old. If you believe a child under 13 has provided personal information through the Extension, please contact us so we can take appropriate action.
11. Sensitive Information
The Extension reads X user bios (profile descriptions) and tweet content, which may contain sensitive personal information (including health conditions, political views, religious beliefs, sexual orientation, or other sensitive self-descriptions). When AI features are used, this content is transmitted to NVIDIA for processing. If you do not want sensitive tweet or bio content transmitted to NVIDIA, do not use the AI reply or AI content filtering features.
12. Changes to This Privacy Policy
We may update this Privacy Policy from time to time. Changes will be posted on this page with an updated "Last updated" date. We encourage you to review this page periodically.
13. Contact Us
If you have questions or concerns about this Privacy Policy or the Extension's data practices, you can contact us on X (Twitter): @ThanksuTrump
隐私权政策
最后更新：2026 年 6 月 29 日
本隐私权政策描述了 Refreshing 浏览器扩展（以下简称"本扩展"、"我们"）如何收集、使用、存储和共享用户数据。本扩展适用于 X（原 Twitter）平台，提供 AI 生成回复、批量取关、关注/屏蔽列表导出、关键词监控自动关注和垃圾账号检测等功能。
安装并使用本扩展即表示您同意本隐私权政策中所述的数据处理实践。
1. 我们收集的数据
1.1 您主动提供的数据
- NVIDIA API 密钥 — 您可以在扩展设置中选择性输入自己的 NVIDIA API 密钥以使用 AI 功能。此密钥存储在
chrome.storage.sync 中，会通过您的 Google 账号在设备间同步。
- 人格列表 — 您可以配置用于回复生成的 AI 人格名列表。存储在
chrome.storage.sync 中。
- 关键词列表 — 您可以配置自动关注监控功能的关键词。存储在
chrome.storage.sync 中。
- 数值设置 — 各种数值偏好（关注延迟、监控间隔、滚动次数、速率限制、浏览时间范围、粉丝阈值等）。存储在
chrome.storage.sync 中。
- 界面语言偏好 — 您选择的扩展界面语言。存储在
chrome.storage.sync 中。
1.2 自动收集的数据
- X Bearer 令牌 — 本扩展通过
webRequest API 拦截 X 自身 GraphQL API 请求中的 Bearer 令牌。此令牌存储在内存和 chrome.storage.session 中（浏览器关闭时清除）。仅用于代表您发起 X 第一方 API 请求（如导出列表、拉黑账号），绝不传输给任何第三方。
- X CSRF 令牌 (ct0) 和用户 ID (twid) — 在 x.com 页面通过
document.cookie 读取。仅用于在您自身会话中认证 X API 请求。仅存储于内存，不写入持久存储，不外传。
- 设备指纹 — 每次页面加载时生成的随机 16 字节十六进制值（非持久标识）。仅在您选择启用 xhunt 排名导出功能时发送给 xhunt 排名服务（见第 6.4 节）。
1.3 从 X (Twitter) 读取的数据
- 推文文本、用户名和图片 — 当您使用 AI 回复功能时，本扩展从页面 DOM 读取推文串内容（文本、作者用户名、图片 base64 编码）。此数据发送至 NVIDIA 进行 AI 处理（见第 6.1 节），扩展不做持久化存储。
- 关注列表和屏蔽列表 — 当您导出关注或屏蔽账号时，本扩展调用 X 第一方 GraphQL API 获取列表。数据包括用户名、显示名、简介、粉丝数和账号元数据。此数据在内存中处理并由您本地保存为 CSV/JSON 文件，不上传至任何扩展运营的服务器。
- 搜索结果（关键词监控） — 关键词监控运行时，本扩展通过 Chrome DevTools Protocol (CDP) 调试器拦截 X 的 SearchTimeline API 响应。提取的数据包括用户名、显示名、用户 ID、简介、粉丝数、认证状态、推文内容和关系状态。此数据存储在
chrome.storage.local 中进行处理。
- 用户资料数据（垃圾检测） — 垃圾检测模块从 X 的 GraphQL 响应和页面 DOM 读取用户资料信息（用户名、用户 ID、简介、显示名、头像、粉丝/关注数、账号创建日期、关系状态）。检测结果缓存在
chrome.storage.local 中。
- 时间线/信息流数据（热点帖子） — 热点帖子功能拦截 X GraphQL 响应以提取推文指标（浏览量、点赞、转推、回复、收藏、全文）用于本地速度计算。此数据不外传。
2. 我们如何收集数据
- webRequest 请求头拦截 — 使用
webRequest 权限从 X API 请求头中捕获 Bearer 令牌。
- fetch/XHR 钩子 — 在页面主世界中包裹
window.fetch 和 XMLHttpRequest，拦截 X GraphQL API 响应用于垃圾检测和热点帖子功能。
- Chrome DevTools Protocol (CDP) 调试器 —
debugger 权限用于：(a) 模拟鼠标移动和点击以执行批量取关和自动关注操作，(b) 关键词监控期间拦截 SearchTimeline 网络响应。当调试器附加时，Chrome 会在页面顶部显示"DevTools 正在调试此浏览器"横幅。调试器不直接读取页面 DOM 内容；DOM 抓取由内容脚本使用标准 DOM API 完成。
- DOM 抓取 — 内容脚本使用
querySelector 和 innerText 从页面 DOM 读取推文文本、用户名、图片和资料信息。
- Cookie 读取 — 读取
document.cookie 获取 X CSRF 令牌和用户 ID 用于认证 X API 请求。
- 用户输入 — 您在扩展弹窗中输入的设置（API 密钥、人格列表、关键词、数值偏好）。
3. 我们如何使用数据
- AI 回复生成 — 推文串内容（文本、用户名、图片）发送至 NVIDIA AI API（经我们的 Cloudflare Worker 代理）以生成建议回复。
- 自动互关（关键词监控） — 监控 X 搜索结果中匹配您关键词和过滤条件的用户，自动关注匹配用户。启用 AI 内容过滤时，候选用户的昵称、简介和推文内容会发送至 NVIDIA 进行评估。
- 批量取关 — 本扩展通过 CDP 模拟鼠标点击 X 的取关确认对话框来自动化取关流程。
- 关注/屏蔽列表导出 — 本扩展从 X API 获取您的关注/屏蔽列表，允许您下载为 CSV/JSON 文件。
- 垃圾账号检测 (MXGA) — 本扩展使用本地启发式规则分析 X 用户资料以识别潜在垃圾账号，可自动拉黑。仅数字 X 用户 ID 发送至公榜查询服务（见第 6.5 节）；不上传简介或推文内容。
- 热点帖子检测 — 本扩展使用本地速度计算监控信息流中快速传播的帖子。不外传任何数据。
- 本地统计展示 — 本扩展维护本地统计数据（关注数、拉黑数、监控统计）用于扩展界面展示，不上传至任何服务器。
4. 数据存储
4.1 chrome.storage.sync（通过 Google 账号跨设备同步）
NVIDIA API 密钥、人格列表、模型 ID、预览数量、关键词监控配置（关注模式、AI 过滤模式、AI 模型、粉丝阈值、滚动次数、延迟、速率限制、浏览时间、面板模式、关键词选择、关键词文本）、界面语言偏好、AI 按钮位置、风险确认标记。
4.2 chrome.storage.local（仅本机持久存储）
关键词监控状态/统计/日志/累计数据/速率队列/拦截的用户数据/会话令牌、批量取关进度和待执行配置、导出面板状态、MXGA 垃圾检测数据（已拉黑 ID 集、拉黑记录、每用户判定缓存、拉黑队列、白名单、跳过状态、统计）、XVM 设置和状态。
4.3 chrome.storage.session（浏览器关闭时清除）
X Bearer 令牌、关键词监控面板窗口 ID、全局 X 写操作时间戳（用于速率限制）。
4.4 IndexedDB（Service Worker 上下文）
人格列表数据库 — 存储人格名、版本号和上次检查时间戳，用于人格更新机制。
4.5 localStorage（在 x.com 页面上）
XVM 面板位置/最小化状态、MXGA 拉黑计时状态、MXGA 引导完成标记。
4.6 仅内存（不持久化）
X CSRF 令牌、X 用户 ID、用于 AI 处理的推文文本/图片、设备指纹、拦截的 GraphQL 响应体（处理后即丢弃）。
5. 数据保留期限
- chrome.storage.sync/local — 持续保留至您卸载扩展或手动清除 Chrome 存储/浏览数据。
- chrome.storage.session — 浏览器关闭时清除。
- MXGA 缓存 — 每用户判定缓存和拉黑记录各上限 5,000 条；自动淘汰最旧条目。
- 关键词监控日志 — 上限 500 条；自动淘汰最旧条目。
- 内存数据 — 页面关闭或 Service Worker 重启时丢失。
6. 与第三方的数据共享
本扩展与以下第三方共享特定数据。我们不出售用户数据。 所有数据共享严格用于功能实现。
6.1 NVIDIA AI API (integrate.api.nvidia.com)
| 触发场景 | 发送数据 | 用途 |
|---|
| AI 回复（经 Cloudflare Worker） | 推文文本、作者用户名、推文图片（base64）、AI 人格名、您的 NVIDIA API 密钥（未配置则使用内置密钥） | 生成 AI 建议回复 |
| 关键词监控 AI 过滤（直连） | 候选用户昵称、简介/描述、推文内容、您的 NVIDIA API 密钥 | 在自动关注前评估用户是否符合内容标准 |
NVIDIA 隐私政策：https://www.nvidia.com/en-us/about-nvidia/privacy-policy/
6.2 Cloudflare Worker (ai-server-worker.ai-refreshing.workers.dev)
我们自行运营的 Cloudflare Worker 作为扩展与 NVIDIA API 之间的代理。当您使用 AI 回复功能时，Worker 接收完整请求载荷（推文文本、用户名、图片、人格名、API 密钥），从 Cloudflare Workers KV 读取人格提示词，然后将请求转发至 NVIDIA。Worker 不记录或存储您的请求数据（除瞬时处理外）。
Cloudflare 隐私政策：https://www.cloudflare.com/privacypolicy/
6.3 DuckDuckGo (lite.duckduckgo.com)
当 AI 回复代理判定需要网络搜索时，本扩展向 DuckDuckGo Lite 发送搜索查询（由 AI 模型根据推文上下文生成，非您的直接输入）。搜索请求中不包含个人信息或 X 账号数据。
DuckDuckGo 隐私政策：https://duckduckgo.com/privacy
6.4 xhunt KOL 排名 (kb.cryptohunt.ai)
当您导出关注或屏蔽列表并勾选"包含 xhunt 排名"选项时，本扩展将 X 用户名列表（公开用户名）、当前页面 URL 和随机会话级设备指纹发送至 xhunt 排名 API 以获取 KOL 排名数据。
| 发送数据 | 用途 |
|---|
| X 用户名列表（来自导出列表）、页面 URL、随机设备指纹、请求签名 | 获取导出账号的 KOL 排名分数 |
6.5 MXGA Edge 服务 (x.zuoluo.tv)
垃圾检测模块与 MXGA Edge 服务通信用于两个目的：
- 公榜查询 — 发送数字 X 用户 ID（每批最多 100 个）查询公榜数据。不发送用户名、简介或推文内容。
- 白名单同步 — 每 6 小时下载维护者维护的已知合法账号白名单。同步过程中不上传用户数据。
6.6 Cloudflare Pages (ai-server-page.pages.dev)
本扩展从 Cloudflare Pages 下载人格列表和版本信息。此为纯下载操作，不上传用户数据。
6.7 X (Twitter) 第一方 API
本扩展使用您的活动会话（Bearer 令牌 + CSRF cookie）向 X 自身 API 端点发起请求，用于以下目的：导出关注/屏蔽列表、拉黑账号（垃圾检测）、触发 Bearer 令牌重新捕获。这些请求作为您自身的认证会话发起，不经过任何扩展运营的服务器。
7. 第三方如何使用您的数据
第 6 节中列出的每个第三方均根据其自身的隐私政策处理数据（各小节已附链接）。我们不控制 NVIDIA、Cloudflare、DuckDuckGo、xhunt 或 MXGA Edge 服务在收到数据后的处理方式。建议您查阅各自的隐私政策。
8. 数据安全
- 所有与外部服务的数据传输均使用 HTTPS/TLS 加密。
- 您的 NVIDIA API 密钥本地存储在
chrome.storage.sync 中，仅在实际使用 AI 功能时传输至 NVIDIA（直连或经 Worker）。
- 当您未配置自己的 API 密钥时，Cloudflare Worker 使用内置兜底密钥池。此密钥池在所有未配置个人密钥的用户间共享。
- 我们不出售、出租或以盈利为目的交换用户数据。
- 本扩展不包含分析、崩溃报告或广告 SDK。
9. 您的选择和控制
- AI 功能 — 您可以不配置 NVIDIA API 密钥。未配置密钥时 AI 回复和 AI 内容过滤功能不会运行。
- 关键词监控 — 您可以不启动关键词监控。可以禁用 AI 内容过滤（将模式设为"关闭"）。
- xhunt 导出 — xhunt 排名功能为可选。如不希望用户名列表发送至 xhunt 服务，请不勾选"包含 xhunt 排名"。
- 垃圾检测 (MXGA) — 您可以在扩展设置中禁用垃圾检测模块。
- 清除数据 — 卸载扩展或前往 Chrome 设置 → 隐私和安全 → 清除浏览数据 → Cookie 及其他网站数据（此操作会清除
chrome.storage）可清除所有扩展数据。关键词监控日志可从扩展弹窗中清除。
- chrome.storage.sync — 通过 Google 账号同步的数据可在 chrome.google.com/sync 管理。
10. 未成年人隐私
本扩展不面向 13 岁以下儿童。我们不会故意收集 13 岁以下儿童的个人信息。X 自身的服务条款要求用户至少年满 13 岁。如果您认为有 13 岁以下儿童通过本扩展提供了个人信息，请联系我们以便采取适当措施。
11. 敏感信息
本扩展会读取 X 用户简介（个人资料描述）和推文内容，其中可能包含敏感个人信息（包括健康状况、政治观点、宗教信仰、性取向或其他敏感自述）。使用 AI 功能时，这些内容会传输至 NVIDIA 进行处理。如您不希望敏感推文或简介内容传输至 NVIDIA，请勿使用 AI 回复或 AI 内容过滤功能。
12. 隐私政策变更
我们可能不时更新本隐私权政策。变更将发布在此页面上，并更新"最后更新"日期。建议您定期查阅此页面。
13. 联系我们
如对本隐私权政策或本扩展的数据处理实践有疑问或顾虑，可通过 X (Twitter) 联系我们：@ThanksuTrump