Last updated: 2026-09-16
SubSensei is a browser extension that overlays synced Japanese subtitles on Crunchyroll (and, experimentally, Netflix), generates furigana readings on your device, and provides a hover dictionary for Japanese learners. This policy describes exactly what data the extension handles and where it goes.
SubSensei now operates a backend service of its own: the shared
subtitle library at subsensei.tanrenworks.com. Earlier
versions of this policy stated that SubSensei had no backend and collected no
data. That is no longer accurate. The extension automatically contributes the
Japanese subtitles it aligns, and automatically reuses subtitles other people
have contributed, so that episodes load quickly for everyone. The sections
below describe precisely what is sent, what is stored, and for how long.
SubSensei does not sell your data or share it with data brokers, requires no SubSensei account or signup, and never sends your Jimaku API key anywhere except Jimaku.
Using the shared library requires no SubSensei account, sign-in or subscription. It has two halves, and both are automatic.
When the extension identifies a supported episode and obtains the platform's subtitle timing, it asks the library whether a matching subtitle is already known. A lookup sends only four things about the episode — the platform, the platform's episode ID, a timing fingerprint of the player's own subtitle track, and the alignment-engine version — plus an installation identifier (below). It does not send the show title or subtitle text. Lookups require no account token.
After the extension performs a new alignment of a Jimaku subtitle — either one downloaded using your Jimaku API key or an eligible file supplied by the shared library for local checking — it sends a compact contribution summary. Every summary includes:
If that exact result is not already known, the extension also uploads the original selected Japanese subtitle, the show or movie display title, episode number and episode title, and Jimaku provenance such as the entry ID and file name. For a season archive, only the selected subtitle file is uploaded, not the whole archive.
Automatic shared-library lookups and contributions have no opt-out setting. Contributions happen in the background and do not delay playback.
A contribution is not generated by subtitle or reference files you load manually, by your own personal sync adjustments, by rejected subtitle candidates, or when confirmed timing is reused from the shared library or your local cache.
The extension generates a random installation identifier (UUID) automatically — it is not tied to your name, email or any account — and sends it with lookups and contributions. The server does not store it as-is: using a server-only secret, it derives an HMAC pseudonym of that identifier and a second pseudonym of the network address that Cloudflare reports for the request (the address itself for IPv4; a normalized /64 prefix for IPv6). SubSensei's application database and subtitle storage keep the pseudonyms, not the raw identifier or address. They are used to de-duplicate evidence and limit abuse, and are not included in stored subtitle files or in responses sent to other users. Cloudflare may separately process or log the network address as the hosting provider, as described under server storage below.
This amounts to pseudonymous collection of episode activity. We state that plainly rather than claiming that no data is collected.
A contribution never includes the page URL, the streaming or video URL, the text or full timing track of the platform's own subtitles, or your Jimaku API key.
| Information or request | Service | Why |
|---|---|---|
| Episode identity, timing fingerprint, installation identifier; contributed subtitles and alignment details | SubSensei shared library (subsensei.tanrenworks.com, hosted on Cloudflare) |
Reuse and contribute aligned subtitles, as described above. |
| The anime series title shown on the page | AniList (graphql.anilist.co) |
Resolve the title to an AniList ID so the right subtitle can be found. |
| AniList ID + episode number + your Jimaku API key | Jimaku (jimaku.cc) |
Download the matching Japanese subtitle file. |
| Japanese word/phrase you hover over | Jisho (jisho.org) |
Return the dictionary definition shown in the popup. |
| Crunchyroll's own subtitle track URL | Crunchyroll CDN | Fetch the platform's timed subtitle used as the alignment reference. |
| Read Netflix's own timed-text (TTML) track and on-page episode metadata | Netflix (www.netflix.com) |
Identify the title/episode and use the platform track as the alignment reference. Netflix support is experimental. |
Lookups, new alignments and dictionary requests originate from activity on a supported watch page. A contribution that could not be delivered immediately is queued locally and may be retried later in the background. Apart from the first row, which is SubSensei's own service, each service is a third party with its own privacy policy governing the request.
Furigana — the hiragana readings shown above kanji — are generated entirely in your browser, using a Japanese morphological dictionary bundled inside the extension (about 18 MB, which is why the download is large). No subtitle text is transmitted to produce them. Subtitle alignment itself also runs locally.
Local extension data is stored in your browser via
chrome.storage.local:
Recent activity diagnostics, such as cache timing and alignment status, are kept in memory for the popup activity log. Enabling debug mode also prints them to the page console. The extension does not send that activity log to the developer; the alignment diagnostics included in contributions are the exception described above.
The extension requests access only to Crunchyroll, Netflix, Jimaku, AniList,
Jisho and its own API host. Release builds pin the exact API host. Unpacked
developer builds may additionally request optional access to
*.workers.dev and localhost so a developer can point
the extension at a test server; these are never used by store builds.
Note that contributions do include alignment diagnostics, so the library does receive technical information about how alignment performed on your device.
Uninstalling the extension removes everything stored on your device — settings, your Jimaku API key, your installation identifier, the pending contribution queue and the subtitle cache. You can also clear the subtitle cache and your API key from the popup and options page at any time.
Uninstalling does not remove data already contributed to the shared library. Contributed subtitle files and the pseudonymous alignment evidence derived from them remain until the maintainer removes them. Because the library has no accounts, there is no self-service deletion page. If you want a contribution or its associated evidence removed, email the address below with enough detail to identify it (for example the show and episode), and I will remove it where it can be identified. Because there are no accounts and identifiers are pseudonymized, it may not be possible to associate a particular stored record with you.
If this policy changes, the "Last updated" date above will change and the revised version will be published on this page, which is the authoritative copy.
For questions about this policy, or to request removal of contributed data, email tim@tanrenworks.com. If you contact this address, the maintainer and email provider receive the email address and message you choose to send and use them to respond to you or handle your request.