Privacy
What we collect on this site
- Waitlist: if you join the waitlist we store the email address you submit and which language version of the page you used, only to email you about the launch. We keep waitlist addresses only until launch announcements have been sent, then delete the list within 90 days — or immediately on request. Nothing else is collected — no analytics, no tracking scripts, no cookies set by us. We do not track visitors across other sites, so we do not respond differently to Do Not Track signals.
- Account pages: the pricing, account and sign-in pages load no analytics and set no cookies of ours by themselves. When you sign in, our API sets one cookie,
abt_session(HttpOnly, Secure, scoped to/api, 30 days), which identifies your session and nothing else. The pricing page loads Paddle's checkout script; Paddle may set its own cookies and local storage on that page and inside the checkout, under Paddle's privacy policy.
Where it goes
Waitlist submissions are stored on servers operated by Aibrify. They are never sold, and never shared with anyone — they sit on server infrastructure we rent and operate ourselves, and access is limited to the developer. Aibrify Translate is built and operated by an independent developer; the data controller for the waitlist is Aibrify, reachable at support@aibrify.com.
Your rights and choices
To be removed from the waitlist at any time, email support@aibrify.com from the address you signed up with. You can also ask what we hold about you (for the waitlist that is only your email address and page language) and have it corrected or deleted. We honor these requests wherever you live; in some places (such as the EU) they are also your legal right. For accounts, the account page shows what we hold that matters day to day (email, plan, usage) and lets you delete it yourself; email us for a full copy, including the name and profile picture we received from Google.
The extension
The extension translates subtitles on YouTube, whole web pages, text you select, and text you type into an input box. None of that traffic goes through a server of ours: we never receive, log or store what you translate. Which translation engine you use decides who else sees the text — the section below spells that out for each of the three.
- No accounts, no analytics, no tracking. The extension works without registration and contains no analytics, telemetry or advertising code. We do not collect, receive, sell or share any personal data from the extension, and there is no cloud sync.
- Where it runs. On YouTube it reads the video’s own subtitle track. Anywhere else it does nothing until you ask: pressing the translate shortcut or clicking the toolbar button lets it work on that one tab, right then. It keeps running on a site only if you turn on “Always translate this site” for that site, or “Selection & hover on all sites” in Settings — both ask your browser for permission at that moment, and both can be revoked in Settings or in your browser’s extensions page at any time.
- Where your text goes when it is translated. There are three engines, and the request always leaves from your own browser — never from a server of ours, which is why we cannot see, log or keep what you translate:
- Chrome’s built-in translator (the free default where your browser supports your language pair): the text is translated on your own computer and does not leave it.
- A free public online engine (currently Google’s public web translation endpoint), used when the built-in one cannot handle your language pair. Your browser contacts that endpoint directly, so the text and your IP address are visible to that provider under its own policies. We are not affiliated with it, we do not proxy the request, and we receive nothing. You can switch this off in Settings and stay on the on-device engine only.
- Your own API key (optional: DeepL, Google, OpenAI, DeepSeek or Gemini). The request goes directly from your browser to the provider you chose, using your key. The key is stored only on your device, is never displayed back to you, and never passes through Aibrify’s servers. That provider’s privacy policy applies to what you send it; you can remove the key and its site permission at any time in Settings.
- Documents. Documents (PDF / Office / text) opened in the extension's viewer stay in your browser. They are never uploaded to us. Paragraph text is sent only to the translation provider you chose (on-device, a public endpoint, your own key, or Pro), exactly as for web pages. Scanned pages have no text layer, so the viewer tells you which pages they are and leaves them untranslated — nothing is sent anywhere for them.
- Translations are not kept. Original and translated text is held in memory only — in the page while you have it open, and briefly in the extension’s background worker, which your browser shuts down on its own when idle. Nothing is written to disk, nothing is uploaded to Aibrify, nothing is synced between your devices, and there is no export or sharing feature for translated text. The one exception is a word you deliberately save, which goes into your local vocabulary book (IndexedDB on your device) along with the sentence and page it came from.
- What stays on your device. Your settings, your per-site “always / never translate” rules, your API key if you added one, a simple counter of how much you have translated (a number of segments and characters — not the text), and your vocabulary book. Uninstalling the extension deletes all of it (this is how Chrome and other Chromium browsers such as Edge work). Export to CSV/Anki is available anytime, free. Review history and the list of words you mark as known stay on this device; they are never uploaded (the same for Free and Pro).
- What the extension requests from us: a small static configuration file (JSON, so we can fix breakage quickly when a site changes its pages) and, if you use word lookup, dictionary packs for your language pair. On first install it opens this site’s welcome page once — an ordinary page visit with no analytics scripts. A dictionary-pack request names the language pair you picked and, like any web request, reaches our host together with your IP address; beyond that we ask for no personal data, and none of these requests carries anything about what you were translating.
The extension’s handling of the data described above adheres to the Chrome Web Store User Data Policy, including its Limited Use requirements. Page and subtitle content is used only to provide the translation and lookup features you asked for — never sold, never used for advertising, and never used for any unrelated purpose.
Accounts and Pro
Everything above stays true for the free features: no account, nothing leaves your browser through us. This section only applies if you choose to sign in — which is needed only for Pro.
- Signing in. When you sign in with Google we receive your email address, name and profile picture from Google and store them with a random account ID. We never see your Google password, and we do not keep Google access tokens beyond the sign-in itself. If you sign up with an email address and password instead, we store the email address and a bcrypt hash of the password — never the password itself.
- Payments. Pro is sold by Paddle.com Market Ltd ("Paddle"), our merchant of record. Paddle collects your payment details, billing address and tax information under its own privacy policy. Paddle's notifications also tell us the amount and currency of each payment so we can confirm a charge; we store only your subscription status, plan, billing period and Paddle customer and subscription references — never card numbers.
- Pro AI features. When you turn on an AI feature (an AI-translated subtitle track, or line-by-line explanations), the subtitle lines being shown at that moment are sent from your browser to our server and forwarded to Google's Gemini API on its paid tier, under Google's data-processing terms, which do not allow using your data to train models. We do not store the original text or the translations. Our server logs record only request metadata — method, path, status, timing and internal event ids — never the text, and never your email.
- AI explanations. When you ask for an explanation of a subtitle line, that line (plus the line before and after it, for context) goes to Gemini the same way. We store only the explanation we generated, under a keyed hash of the sentence (a pseudonymised key, so the next person who asks about the same line gets the cached answer instead of a new model call). A stored explanation may quote at most the word you asked about and one clause of the line; we never store the full line, and we do not store the explanations you read as part of your account. Entries that have not been used for 180 days are deleted, and we delete an entry on request.
- Usage counters. To enforce the fair-use quota we keep, per billing period, how many subtitle lines and AI explanations you have used, with timestamps. Counters for the current and the previous period are kept; older ones are deleted automatically within about a month.
- Credits. If you buy credits, we keep a ledger of each purchase, bonus, hold, charge, release and refund: the pack, the amount and currency Paddle reported, the number of credits, a job ID and a timestamp — never the file name or its contents. The ledger stays as long as your account exists and is erased with it 30 days after deletion.
- Emails. We send service emails about your account and subscription: to confirm your email address when you sign up, to reset your password when you ask us to, to tell you when your password was changed, when your quota reaches 80 % and 100 % of the fair-use allowance, and before any change to Pro terms (at least 30 days ahead — a commitment we honor by hand, not by automation). If we ever have to suspend an account for abuse, we email you before doing so — also by hand. Paddle, as merchant of record, emails you about payments: receipts, failed payments and card updates. We do not send marketing email.
- Deleting your account. Settings → Account in the extension, or the account page on this site, deactivates your account immediately. If you sign in again to the same account (with Google or with your email and password) within 30 days, it is restored exactly as it was. Otherwise, after 30 days, we erase the account together with its subscription record and usage counters. Cancel any subscription first (Paddle keeps its own transaction records as required by tax law).
- Where it lives, how long. Account data is stored on servers we rent and operate ourselves, for as long as your account exists. Google (sign-in, Gemini) and Paddle may process data outside your country; for transfers from the EU/UK we rely on the EU–US Data Privacy Framework where the recipient is certified, and otherwise on Standard Contractual Clauses. Service emails are delivered through Resend, an email delivery provider, which receives your email address and the message text. Internal alerts about payments reach our team's Slack with your email address masked.
- EU and UK representative. We have not yet designated a representative under Article 27 GDPR / UK GDPR. Until we do, please contact support@aibrify.com for anything data-related; we answer every request ourselves.
Children
Neither this site nor the extension is directed at children under 13, and we do not knowingly collect personal data from them — the extension collects no personal data from anyone by design.
Changes to this policy
Any change will be posted here with a new “updated” date at the top. If a change materially affects how waitlist data is handled, we will email the waitlist before it takes effect. Pro has its own section above; it will be updated before any change in what is sent where.
中文摘要:本站只在你加入 waitlist 时保存你提交的邮箱与页面语言,仅用于上线通知,上线通知发出后 90 天内删除(也可随时要求立即删除);不使用统计脚本、不设 Cookie、不出售或共享数据。你可随时要求查询、更正或删除你的 waitlist 数据,邮箱 support@aibrify.com。
扩展:无账号、无统计、无广告代码、无云同步,不收集任何个人数据。在哪里运行:YouTube 上读视频自带的字幕轨;其他网站默认什么都不做——按快捷键或点工具栏按钮,才对当前这一个标签页生效;只有你为某个网站打开「总是翻译本站」,或在设置里打开「所有网站的划词与悬停」,它才会常驻,这两项都在你操作的那一刻才向浏览器申请权限,也随时可以在设置或浏览器的扩展页里撤销。文字去哪儿:三种引擎,请求一律从你自己的浏览器发出、不经我们的服务器——所以我们看不到、也记不下、更不会保存你翻译了什么;至于还有谁能看到这段文字,取决于你用的是哪个引擎:① Chrome 内置翻译器在你的电脑上完成,文字不出设备;② 免费公开在线引擎(当前是 Google 网页版翻译端点),用于内置引擎覆盖不到的语向,由你的浏览器直连,因此该提供方会看到这段文字与你的 IP,适用其自己的政策,我们与其无隶属关系、不做中转、不接收任何内容,你也可以在设置里关掉它、只用端侧引擎;③ 你自己的 API key(DeepL、Google、OpenAI、DeepSeek、Gemini 之一,可选),用你的 key 从你的浏览器直发你选的提供方,key 只存本机、不回显、不经我方,适用该提供方的隐私政策,随时可在设置里删除 key 与对应站点权限。文档:在扩展查看器里打开的文档(PDF / Office / 文本)留在你的浏览器里,不会上传给我们;只有段落文本会发给你选择的翻译提供方(端侧引擎、公开端点、你自己的 key 或 Pro 引擎),规则与网页翻译完全一致。扫描页没有文字层,查看器只会告诉你哪几页是扫描件、不做翻译,也不会为它们向任何地方发送内容。译文不保留:原文与译文只存在于内存——页面打开期间在页面里,另有一份短暂留在扩展的后台工作线程里,浏览器闲置时会自行把它关掉;不落盘、不上传、不在设备间同步,也没有针对译文的导出或分享功能;唯一的例外是你主动存下的生词,连同来源句子与页面存进本机生词本。存在本机的东西:设置、每个站点的「总是 / 从不翻译」规则、你自己填的 API key、一个「已翻多少」的计数(段数与字符数,不含文字本身)、生词本;卸载扩展即全部删除,导出 CSV / Anki 永远免费。复习记录与已认识词只存本机,不上传(免费与 Pro 相同)。扩展向 Aibrify 请求的只有静态配置文件与词典包:这类请求会带上你选的语言对,并且和任何网络请求一样,会让我们的主机看到你的 IP;我们不索取任何个人数据,请求里也不含你翻译的内容。首次安装会打开一次本站欢迎页。页面与字幕内容只用于你要求的翻译与查词功能,不出售、不用于广告、不用于任何无关目的。
本站与扩展均不面向 13 岁以下儿童,也不会有意收集其数据。政策变更会在本页发布并更新页顶日期,重大影响 waitlist 数据的变更会先邮件通知;付费功能见上文「账号与 Pro」段。本站不做跨站追踪,因此对 Do Not Track 信号不作区别处理。
账号与 Pro:免费功能一切照旧、无账号。只有为了 Pro 才需要登录(Google 或邮箱 + 密码):用 Google 登录时我们从 Google 收到并保存你的邮箱、姓名、头像与一个随机账号 ID,看不到你的 Google 密码,不保留 Google 令牌;用邮箱注册时我们保存邮箱与密码的 bcrypt 哈希,永远不存明文。付款由我们的记录商 Paddle 处理(银行卡信息只经 Paddle,我们只收到订阅状态、套餐、周期与 Paddle 的客户 / 订阅编号);Paddle 的通知会告诉我们付款金额与币种,用于核对扣款。你开启 AI 功能时,当时显示的字幕行会经我们的服务器转发给 Google Gemini API 付费档(受 Google 数据处理条款约束、不用于训练);我们不保存原文与译文,服务器日志只记方法、路径、状态、耗时与内部事件编号。你点「AI 讲解」时,那一行(及前后各一行作语境)同样经我们转发给 Gemini;我们只保存我们生成的讲解文本,以句子的密钥哈希(假名化)作键,供下一位问同一句的人复用——讲解最多引用你问的词和一个从句,绝不保存整句;180 天没人用到即删除,收到要求即删。为执行公平使用额度,我们按计费周期保存已用行数 / 讲解次数与时间戳,只留当前与上一周期。只发服务邮件:邮箱确认、密码重置、密码已更改、额度 80% / 100%、Pro 条款变更(提前 30 天,人工执行);付款收据与扣款失败由 Paddle 发,不发营销邮件;若因滥用需停用账号,会先邮件通知。删除账号:立即停用;30 天内用同一账号(任一登录方式)再登录即恢复;否则 30 天后连同订阅记录与用量清除。跨境传输依 EU–US 数据隐私框架或标准合同条款。服务邮件经 Resend 投递(它会收到你的邮箱与邮件内容);内部付款告警发到我们的 Slack,邮箱已打码。GDPR 第 27 条的 EU / UK 代表暂未指定,数据相关请求请发 support@aibrify.com。英文版为准。
点数:若你购买点数,我们保存每笔购买、附赠、冻结、扣除、释放与退款的账本记录(包、Paddle 报告的金额与币种、点数、任务 id、时间),不含文件名与文件内容;账本随账号保留,账号删除 30 天后一并清除。