VoxAI 隐私政策
最后更新：2026 年 5 月 5 日 · 适用版本：v1.0
VoxAI 是什么
VoxAI 是一款 macOS 上的语音输入工具，主要用途是把你说出的中文实时转写成文字，自动复制到剪贴板，让你能粘贴到 Claude Code、Cursor 等 AI 编程助手里使用。
VoxAI 不做什么
- 不收集任何个人信息。VoxAI 不要求注册、不需要登录、没有用户账号系统。
- 不上传你的音频或转录文字到 VoxAI 的服务器—— v1.0 根本没有"VoxAI 的服务器"。
- 不追踪用户，不集成任何第三方分析或广告 SDK。
- 不收集使用统计 / 崩溃日志 / telemetry。
- 不读取你的文件 / 通讯录 / 日历 / 浏览器数据。
VoxAI 实际访问哪些数据
麦克风
VoxAI 需要麦克风权限来捕获你的语音。捕获到的音频仅在内存中用于发送给 Apple 系统语音识别框架，不写入磁盘、不上传到任何 VoxAI 控制的服务器。
语音识别（Apple SFSpeechRecognizer）
VoxAI 使用 Apple 提供的 SFSpeechRecognizer 系统框架做语音转文字。根据 Apple 的隐私政策，音频数据可能会上传到 Apple 的服务器进行识别处理。这是 macOS 平台上语音识别的标准行为，VoxAI 选择默认服务器模式以提供最佳识别准确率。
关于 Apple 如何处理这些数据，请参考 Apple 官方隐私政策。
系统剪贴板
录音停止后，VoxAI 会把转录文字自动写入系统剪贴板（默认开启）。剪贴板的内容由你完全控制——你可以在设置里关闭这个行为，也可以在粘贴前自行清除。
UserDefaults（应用偏好设置）
VoxAI 在你这台 Mac 的应用专属 UserDefaults 容器里存储一项配置：是否启用"录音停止后自动复制到剪贴板"。这项配置不会同步到任何地方。
VoxAI 不联网
VoxAI v1.0 的 entitlements 只包含两条权限：
com.apple.security.app-sandbox（App Store 强制要求）com.apple.security.device.audio-input（麦克风访问）
没有 network.client、没有 network.server。VoxAI v1.0 在技术层面就无法发起任何外部网络连接。
儿童使用
VoxAI 不专门面向 13 岁以下儿童。VoxAI 不收集任何数据，因此也不会收集儿童的数据。
政策变更
如果未来版本（v1.1+）引入需要联网的功能（如 Cloud TTS、MCP server 等），本政策会同步更新，并在更新版本里通过应用内提示告知用户。
联系
如有疑问，请通过 GitHub Issues 联系作者。
VoxAI Privacy Policy
Last updated: May 5, 2026 · Applies to: v1.0
What VoxAI is
VoxAI is a macOS dictation tool. It transcribes your speech in real time, auto-copies the result to your clipboard, and lets you paste into AI coding assistants like Claude Code or Cursor.
What VoxAI does NOT do
- Does not collect any personal information. No registration, no login, no user accounts.
- Does not upload your audio or transcripts to any VoxAI server — v1.0 has no "VoxAI server" to upload to.
- Does not track users. No third-party analytics, no advertising SDKs.
- Does not collect usage stats, crash logs, or telemetry.
- Does not read your files, contacts, calendar, or browser data.
What data VoxAI actually accesses
Microphone
VoxAI requires microphone access to capture your speech. Captured audio is held in memory only and forwarded to Apple's Speech Recognition framework. It is never written to disk or sent to any server controlled by VoxAI.
Speech Recognition (Apple SFSpeechRecognizer)
VoxAI uses Apple's SFSpeechRecognizer system framework. Per Apple's privacy policy, audio may be transmitted to Apple's servers for recognition processing. This is the standard behavior for speech recognition on macOS. VoxAI keeps the default server mode for best accuracy.
Apple's data handling is governed by their Privacy Policy.
System clipboard
When you stop recording, VoxAI writes the transcript to the system clipboard (enabled by default). The clipboard's contents are entirely under your control — you can disable this in Settings, and you can clear the clipboard manually before pasting.
UserDefaults (app preferences)
VoxAI stores one preference in this app's private UserDefaults container on your Mac: whether auto-copy on stop is enabled. This is not synced anywhere.
VoxAI is offline
VoxAI v1.0's entitlements include only two permissions:
com.apple.security.app-sandbox(required by the App Store)com.apple.security.device.audio-input(microphone access)
There is no network.client, no network.server. VoxAI v1.0 is technically incapable of making any outbound network connection.
Children
VoxAI is not directed at children under 13. VoxAI does not collect any data, so it does not collect children's data.
Policy changes
If future versions (v1.1+) introduce features that require networking (such as Cloud TTS or an MCP server), this policy will be updated and users will be notified inside the app at update time.
Contact
For questions, contact the author via GitHub Issues.