Privacy Policy

Last updated: September 22, 2026

This Privacy Policy explains how Hudsonly LLC (“we”, “us”) collects, uses, and protects information when you use ShadowEars (the “Service”).

Information we collect

Account data — your name, email, and authentication details.
Meeting content — the audio you record or upload, the notes and photos you add to a recording, and the transcripts, summaries and action items generated from it.
Questions and calls — the questions you type or speak in Ask and the answers to them, and, during a voice call, your voice.
Voiceprints (optional) — if you choose to enroll a speaker, a numeric voice embedding used to recognize them in your workspace’s meetings. See “Voiceprints (biometric data)” below.
Usage data — transcription minutes, feature usage, and diagnostic logs.
Billing data — subscription and payment status. Card details are handled by Stripe; we do not store full card numbers.

How we collect it

From you, directly. Account details when you sign up; recordings you make in the mobile app, the browser, the desktop app or the Chrome extension, which are uploaded to us; files you import; notes, photos and questions you add.
Generated from your content. Transcripts, summaries, action items, search indexes and voice identification are produced from your recordings as described below.
From tools you connect. Recordings imported from services you link, and the account access described under “Connections”.
Automatically. Usage and diagnostic data as you use the Service, and the analytics described under “Cookies & analytics”.

How we use information

to transcribe your recordings into text and work out who said what;
to write summaries and pick out action items from your transcripts;
to index your transcripts so you can search them and ask questions about them;
to answer the questions you ask in Ask, in writing or in a voice call;
to draft follow-ups (such as an email or a calendar event) that you then review;
to describe photos and video stills taken during a recording, so they can be searched;
to provide MCP access to your meetings from AI assistants you connect;
to operate billing and enforce plan limits and usage-based overage;
to secure, maintain, and improve the Service; and
to communicate with you about your account and support requests.

We do not sell your personal information or your meeting content.

AI providers

Transcription, summaries, search, Ask and voice calls are done by AI services. Some of them are run by other companies, so to provide these features we send your meeting content to them:

Transcription — the audio of your recordings is transcribed on our own servers or by Modal, a cloud computing company that runs our speech-to-text model for us.
Summaries, action items, search, Ask answers, follow-up drafts and photo descriptions — the relevant transcripts, summaries, notes, photos and your questions are sent to OpenRouter, which passes them to the AI model that does the work. Those models include ones from OpenAI and Google.
Voice calls — your voice during a call is streamed to Google’s Gemini, which answers out loud.

We send this content only so the provider can perform that task and return the result to us — a transcript, a summary, an answer or a spoken reply. We do not sell it. These providers are required to protect your content with the same or equal protection as this policy provides.

We ask first. Before anything is sent, the app shows you what is sent, who receives it and why, and asks for your permission. If you decline, you can still record and your recordings are kept, but they are not transcribed or summarized, and Ask and voice calls stay off. You can turn AI features on or off at any time — in the mobile app under Settings, and on the web under Workspace settings. Your choice is recorded with the date and the version of the explanation you were shown.

Subprocessors

We rely on a small number of service providers to run ShadowEars:

Google Cloud — hosting and object storage for recordings.
Stripe — subscription billing and payments.
Modal — computing that runs our speech-to-text model to transcribe recordings.
OpenRouter — access to the AI language models (including models from OpenAI and Google) that write summaries, index transcripts for search, answer questions, draft follow-ups and describe photos.
Google (Gemini) — the live voice model behind voice calls.
If a ShadowEars administrator configures a custom transcription or summarization provider for your workspace, your content is processed by that provider under its terms instead.

Your meeting content and consent

You control what you record. You are responsible for obtaining the consent of meeting participants where required by law. You can delete meetings from your library; deletion removes the associated recording and transcript from active storage.

Share links

You can create a link that lets anyone who opens it read a single meeting's summary without signing in. Sharing is off by default and is turned on per meeting by someone with write access to the workspace.

By default a share link exposes only the meeting's title, date, duration and summary. You can additionally choose, per meeting, to include the transcript and the audio recording; both are off unless you turn them on. Action items, snapshots, your workspace name and the identities of other members are never exposed by a share link.

Anyone holding the link can open it, so treat a shared meeting as public — and remember that a recording contains other people's voices, which is why publishing the audio asks you to confirm. You can untick either option, or turn the link off entirely, at any time from the meeting page; the change applies to the existing link immediately. Turning a link off stops it working for everyone, and sharing again creates a different link that starts from summary-only. Shared pages are marked noindex so search engines do not list them.

One thing worth knowing before you paste a link somewhere: chat apps such as Slack and WhatsApp fetch the page to draw a preview card, and that card shows the meeting's title. The summary, transcript and audio are never part of a preview — but the title is, and everyone in that conversation can see it without opening the link.

We count how many times a shared link has been opened, so you can see whether anyone read it. That count is a number and a date stored on the meeting — we do not record who opened it, or from where, and there is no tracking script on a shared page. Turning the link off resets the count to zero.

Connections (Gmail, Google Calendar, and other tools)

You can optionally link outside tools — such as Gmail or Google Calendar — so ShadowEars can turn a meeting follow-up into a real action (for example, drafting an email or a calendar event). These connections are per project and entirely opt-in:

What we store. When you link a tool, we store the OAuth access and refresh tokens, encrypted at rest, so we can perform actions on your behalf. You can unlink a connection at any time, which revokes our stored access.
What we do with it. We use the connected account only to perform the specific actions you explicitly approve (for example, sending an email you have reviewed, or creating a calendar event). We do not read your inbox or calendar beyond what those actions require, and we never act without your approval.
Google Limited Use. ShadowEars’ use of information received from Google APIs adheres to the Google API Services User Data Policy, including its Limited Use requirements. We do not use Google user data for advertising, we do not sell it, we do not transfer it to third parties except as needed to provide the feature you enabled or as required by law, and we do not allow humans to read it except with your explicit consent, for security, or as required by law.

Voiceprints (biometric data)

ShadowEars can recognize a person’s voice across meetings so their name is filled in automatically. To do this, we derive a mathematical “voiceprint” (a numeric embedding) from a speaker’s audio in a meeting you record. A voiceprint is biometric data, so it is handled with additional care:

Opt-in only. No voiceprint is created unless a member of your workspace explicitly names a speaker and confirms they have that person’s consent to store it. Voice recognition is off until you enroll someone.
Purpose. Voiceprints are used solely to label speakers in your own workspace’s transcripts. They are never used to identify people across workspaces, and are never sold or shared.
Scope. A voiceprint is stored per workspace and matched only against meetings in that workspace.
Storage. Voiceprints are encrypted at rest.
Deletion. You can delete any enrolled voice at any time from the “Known voices” screen; deletion stops all future matching. Voiceprints are also removed when the workspace is deleted.
Your responsibility. You must obtain any consent required in your jurisdiction before enrolling a person’s voice. Some laws (for example BIPA in Illinois, or the GDPR) treat voiceprints as sensitive biometric data with specific consent requirements.

Data retention

We retain your content for as long as your account is active or as needed to provide the Service. When you delete content or close your account, we remove it within a reasonable period, except where retention is required for legal or security reasons.

Security

Data is transmitted over TLS and stored with access controls. Projects are access-controlled spaces. No method of storage or transmission is perfectly secure, but we work to protect your information.

Your rights

Depending on your location, you may have rights to access, correct, export, or delete your personal information.

You can export your data yourself at any time, on every plan: sign in, open Workspace settings, and use Download my data under “Your data”. The file is JSON and covers your profile, the workspaces you belong to, and the recordings you made, with their transcripts and summaries. You can also delete your account yourself.

For anything else — correcting information, or a question about what we hold — contact us at contact@shadowears.com.

Cookies & analytics

We use privacy-friendly analytics (Google Analytics 4) to understand how the site and app are used so we can improve them. When we run ad campaigns we also use advertising measurement tags (Google Ads and TikTok) to see which ads led to a signup. Analytics and any advertising cookies stay off by default — nothing is set until you choose “Accept” in the consent banner (we use Google Consent Mode, and the TikTok tag is not even loaded unless you accept). We don’t sell your data. You can decline, or change your mind later by clearing the site’s storage. Analytics never includes your meeting content.

Separately, we record which campaign or link brought you to our site — the campaign tags and ad click IDs that the link itself carries (utm_*, gclid, and similar), the referring site, and the page you landed on. This is stored in your browser and sent with your signup so we know which channels are worth paying for. It uses no cookies, is never shared with advertisers, and contains no information about you personally.

Changes

We may update this policy. Material changes will be communicated through the Service or by email.

Contact

Questions about privacy? Contact us or email contact@shadowears.com.