Privacy Policy

Last updated: 20 September 2026

Toldmark ("we", "us") provides this service. This policy explains what personal data we collect, why we collect it, how long we keep it, and what rights you have over it.

If you have any questions about this policy or want to exercise any of your rights, email privacy@toldmark.com.

Toldmark makes a YouTube creator's video archive searchable. There are two kinds of people who use it, and we treat their data differently:

• Creators, who create an account, connect their YouTube channel, and build an archive.
• Viewers, who follow a link a creator has shared and ask questions about that archive. Viewers do not need an account.

1. Data we collect from creators

Account information

When you register, we store your email address, your username and display name, and a cryptographic hash of your password. We never store your password itself. If you upload a profile picture, we store the image you uploaded.

YouTube connection

When you connect your YouTube channel, Google asks you to authorise Toldmark. We request a single permission scope (youtube.force-ssl), which is what the YouTube Data API requires in order to read your channel's caption tracks.

Google's own consent screen describes this permission broadly: it lists the ability to view, edit, and permanently delete your videos, ratings, comments, and captions. We ask for it because it is the narrowest scope Google offers that can read caption tracks; the read-only scope cannot. Toldmark does not use the write side of this permission. Every request we make to YouTube goes through a single checkpoint in our code that only performs read operations, and automated tests fail if anything in the codebase tries to bypass it or send a write request.

From that authorisation we store:

• Access and refresh tokens, encrypted with AES-256-GCM
• Your channel ID, channel title, and uploads playlist ID

We do not request or receive your Google account email address, your name, or your Google profile picture. Connecting YouTube is not a way to sign in to Toldmark; it only links a channel to an account you already have.

You can disconnect your channel at any time, either from within Toldmark or from your Google Account permissions page. Disconnecting marks your channel as disconnected. The stored tokens and the archive built from it are deleted after 7 days, or straight away if you choose “Delete forever now”. If a deletion cannot be completed at that moment, we show that it is still in progress and finish it on our next clean-up.

Video content

For each video you add to an archive, we retrieve and store:

• The caption track (transcript), obtained through YouTube's official captions API
• Video metadata such as title, video ID, and publication date
• Comments on the video, where you have chosen to include them
• A processed version of the transcript, split into segments and converted into numerical representations ("embeddings") that make semantic search possible

Transcripts obtained through the YouTube API are refreshed on a rolling 30-day cycle, as YouTube's Developer Policies require. If a video is deleted or made private on YouTube, the corresponding transcript, segments, embeddings, and search index entries are removed from Toldmark.

Answers, conversations and published pages that already cited the video keep the citation's title, publication date and link to YouTube.

Data about other people

Comments retrieved from your videos include the text of the comment and the display name of the person who wrote it. These people are not Toldmark users and have not agreed to anything with us.

We handle this data narrowly. Comments are only retrieved when you explicitly request a comment analysis — never automatically. They are only ever shown to you, the creator. They are never included in answers given to viewers of a shared archive, and they are never used for any purpose other than producing the analysis you asked for.

The lawful basis for this is legitimate interest: analysing responses to your own videos is a normal part of running a channel, and the data involved is already public on YouTube. If a commenter asks us to remove their data, email us and we will do so.

When a video is removed from Toldmark, its comments are deleted along with everything else.

Conversations

When you ask a question, we store the question, the answer, and the sources it cites. For each source we keep the video's title, its publication date and a link to the moment in the video — not the transcript passage itself. This lets you return to earlier conversations and lets the system refer back to what was already discussed.

For every answer we also keep a technical record of how it was produced, so that we can find out why an answer was wrong. It contains the wording of the question and how our system rewrote it, the search steps taken, the identifiers and ranking scores of the passages considered, a short irreversible fingerprint (hash) of each passage's text, the identifiers of the earlier messages used as context, and timings. It does not contain the transcript text, and it does not contain your name, email address, account identifier or browser identifier. It is deleted together with the conversation it belongs to. It is not shown to creators or viewers, and we use it only to investigate and improve how Toldmark answers.

You can rate an answer with a thumbs up or down. A thumbs down asks for a reason from a list, or a short note in your own words. We store the rating, the reason, the note, and technical details about the answer (such as the type of question and how long it took). The rating record holds no name, email address or account identifier. Ratings are not shown to creators and are deleted together with the answer they belong to.

After a thumbs down you may also choose to share this question and answer with us. If you choose "Yes", we mark the answer so that we can look at this question and answer when investigating problems. You can withdraw it at any time by emailing privacy@toldmark.com; we will remove the shared mark within 7 days. If you asked as a visitor without an account, tell us the question you asked and roughly when, so that we can find it.

Conversation text and the technical records are stored unencrypted. We want to be direct about this rather than leave it unsaid. Encrypting it at rest would mean decrypting it on every read, and conversations are read constantly — each new message needs the earlier ones for context. On the infrastructure Toldmark currently runs on, that overhead is significant. The data is protected by access controls and by the security of the hosting environment, but not by encryption at rest. If this matters to you, do not put sensitive information into conversations.

Usage records

We keep an internal record of each AI request: which workspace it belonged to, which category of usage it fell under, how many tokens it consumed, and what it cost. These records exist so we can control costs and enforce spending limits. They do not contain the content of your questions or answers.

These records also note which conversation and which turn within it the request belonged to, so we can see how cost grows over a conversation. They still contain no part of your questions or answers, and no identifier that points to you.

2. Data we collect from viewers

If you open an archive through a link a creator shared with you, you do not need an account and we do not ask for your name or email address.

We place a single cookie in your browser containing a randomly generated identifier. This identifier lets you see your own previous questions in that archive and nobody else's. It contains no personal information. On our servers we do not store the identifier itself — only a one-way cryptographic hash of it.

We store the questions you ask and the answers you receive, together with the technical record and any ratings described in section 1. These are deleted after 12 months.

If a creator publishes a question and answer on their archive page, that published copy is separate from your conversation and stays until the creator removes it.

The creator whose archive you are using can see the questions asked of their archive and the answers given, including where several questions form a single conversation. This is how they decide which questions are worth publishing as curated answers on their archive page.

They cannot see who asked. Nothing they see carries your name or email address, and nothing connects one conversation to another or to you. If you sign in, we link the conversations you started earlier in the same browser to your account so you can find them again; that link is visible to us, not to creators.

3. Cookies

Toldmark uses only first-party cookies. We do not use advertising cookies, tracking pixels, or third-party analytics services.

All of these are strictly necessary for the service to work.

We also record product analytics in our own database: which features are used, together with the conversation and answer an event concerns and, for signed-in users, the account, or for visitors a one-way hash of their browser identifier. These records are deleted after 12 months, and with the account when an account is deleted. They are not shared with any third party.

4. Who else processes your data

We use a small number of service providers. Each one processes data only to deliver the service.

OpenAI — When you ask a question, the relevant portions of the transcript, your question, and the recent conversation are sent to OpenAI's API to generate an answer. Transcripts are also sent to generate embeddings when a video is first added. When you request a comment analysis, the comments and their authors' display names are also sent to OpenAI.

We send these requests with storage disabled, so the content is not retained in OpenAI's systems for later retrieval. OpenAI does not use data submitted through its API to train its models. OpenAI applies its own limited retention period for abuse monitoring, governed by its terms rather than by us.

We do not send your name, email address, or account identifier to OpenAI. A randomly generated workspace identifier is included to improve response caching.

Google (YouTube Data API) — We use the YouTube Data API to retrieve captions, video metadata, and comments from the channel you have connected. Your use of Toldmark is subject to the YouTube Terms of Service, and Google's handling of your data is governed by the Google Privacy Policy. You can revoke Toldmark's access to your Google account at any time at Google's security settings.

DigitalOcean — Provides our database and hosting infrastructure (servers in Toronto, Canada).

Resend — Sends the account emails (email verification and password reset). The provider is based in the United States and delivers through Amazon Web Services (Amazon SES) in the United States. Each email contains your email address and a one-time link; it does not contain your chats, questions, answers, or video content.

5. Where your data is stored

Toldmark stores data on servers located in Canada or the European Union.

Account emails are the exception: they are sent through a United States-based email provider (see section 4).

If you are in the United Kingdom or the European Economic Area and your data is stored in Canada, that transfer relies on the European Commission's adequacy decision for Canada, which recognises Canadian data protection law as providing an equivalent level of protection.

6. How long we keep things

When you delete your account from your profile page, we begin a 30-day period during which you can reverse it from your account page or by contacting us. After 30 days your account and everything tied to it — workspaces, transcripts, embeddings, conversations, ratings, the technical records of your answers, YouTube tokens and analytics about your use — is permanently deleted. A question you asked in someone else's archive that its creator already published stays on their page until they remove it.

If you ask us to delete your data by email rather than through the product, we will complete the deletion within 7 days of receiving your request.

7. Legal basis for processing

If you are in the United Kingdom or the European Economic Area, we process your data on the following bases:

• Contract — to provide the service you signed up for: your account, your archive, and your conversations.
• Legitimate interests — to keep the service secure, prevent abuse, control operating costs, and understand which features are used. We have considered these against your rights and consider them proportionate.
• Consent — for connecting your YouTube channel. You give this consent through Google's authorisation screen, and you can withdraw it at any time by disconnecting the channel.

8. Your rights

If you are in the United Kingdom or the European Economic Area, you have the right to:

• Access your data and receive a copy of it
• Correct data that is inaccurate
• Delete your data
• Restrict or object to how we process it
• Withdraw consent where processing is based on consent
• Receive your data in a portable format

You can delete your account yourself from your profile page. For anything else, email privacy@toldmark.com. We do not currently offer a self-service data export; if you ask for a copy of your data, we will prepare and send it to you.

We will respond to any request within 30 days.

If you are in the UK you have the right to complain to the Information Commissioner's Office. If you are in the EEA you may complain to your national data protection authority.

If you are in Canada, your rights under PIPEDA and British Columbia's Personal Information Protection Act are broadly equivalent. You may complain to the Office of the Privacy Commissioner of Canada, or, since we are based in British Columbia, to the Office of the Information and Privacy Commissioner for British Columbia.

9. Security

Passwords are stored as hashes and never in plain text. YouTube access and refresh tokens are encrypted with AES-256-GCM. Traffic between your browser and Toldmark is encrypted in transit.

As noted in section 1, conversation text is not encrypted at rest.

Our own server logs record technical details such as a workspace or video identifier, a request's type, and how long it took — never the text of a question or an answer.

Toldmark is an early-stage product operated by a small team. We take security seriously, but we do not hold formal certifications such as SOC 2 or ISO 27001, and we think you should know that.

10. Children

Toldmark is not intended for anyone under 16, and we do not knowingly collect data from children. If you believe a child has provided us with personal data, email us and we will delete it.

11. Changes to this policy

If we make significant changes, we will update the date at the top of this page and notify account holders by email before the changes take effect.

12. Contact

Toldmark
privacy@toldmark.com