Privacy Policy
SourceSecure exists to stop client details reaching an AI tool. Collecting those details ourselves would make it a poor product, so we do not.
Effective 11 November 2026 · version privacy-2026-11-11
Effective 11 November 2026 (version privacy-2026-11-11). This policy explains what JOLIVE Labs Inc. ("SourceSecure", "we") collects when you use SourceSecure, why, where it goes and how long we keep it. Our privacy officer is Jacob Oliveira, 95 King William Court, Cambridge, Ontario N3C 4J3, support@sourcesecure.ca.
1. The short version
- What you type and attach on AI chat sites is checked on your own computer. We never receive your files, and we receive no message text until it has been through redaction there.
- We never store your prompts on our servers. Your vault (on a Guard Enterprise seat, also your notes, files and short summaries of your AI chats and meetings), with your own list of names to catch and your list of things marked not sensitive, is encrypted on your computer, and an encrypted copy is kept on your SourceSecure account, so it comes back if you remove and add the extension again and is the same on every computer you sign in on. It is never used to train AI models.
- When you ask your vault to learn from a document, a copy with names and client details replaced by stand-ins is read once by our server, which keeps nothing. The document itself is never sent to be read; it and what was learned are kept in your vault.
- For the optional AI features, text is sent to us and to OpenAI with the names and numbers found in it already replaced by stand-ins. We do not store it, and it is sent to OpenAI with OpenAI's storage option switched off. What OpenAI may keep is set out in 4.2.
- On a Guard Enterprise firm, short facts about the firm's work and the names of its clients that your vault learns are shared with your firm's team vault (see 5.3). Facts about you or your own life are never shared.
- While SourceSecure is on, an AI site's own microphone is held. Dictation is done by SourceSecure on your computer, and your voice is never recorded, kept or sent to us. Voice mode, which sends your voice straight to the AI site, is let through only once at a time when you choose, unless your firm has turned it off.
- Your firm's admin sees what kind of detail was caught and a masked form of it (such as the last four digits of an account number), and when you paused protection, switched it off, or set it to check only when you ask (manual) or not at all; never your messages.
- Our database is in Canada (Montreal). We do not sell information or use it for advertising or to train AI models.
2. Who this policy is for
SourceSecure is sold to businesses ("firms"). The firm decides who gets a seat and which features are on, and is responsible for its clients' information. We handle that information on the firm's behalf under a Data Processing Agreement. This policy also covers visitors to sourcesecure.ca.
3. What we collect, and why
| What | Why | How long |
|---|---|---|
| Your email address, your team and seat type | To sign you in and give you the features your firm bought | Until you are removed from the team, or the firm's records are deleted (30 days after its subscription ends, or at once if its owner deletes it) |
| Sign-in records (through Google Firebase Authentication) | To keep your account secure | As Firebase keeps them |
| For each detail caught: when, which AI site, the kind of detail, what you chose to do (or, if checking was set to manual, that it went without a check), and a masked form made on your computer (e.g. •••• 4242 or Rob••• Ch••). The masked form keeps only the last few digits of a number; the first half of each word of a name, an address or other words; the start of an email's name and of its domain, with its ending (e.g. jane••••@smit••••.ca); or the first part of a postal or ZIP code (e.g. N3C •••) | So your firm can see the tool is working and where client details come up | 90 days from when it is received |
| When protection was paused or switched off, or checking was set to manual (only when you ask) or off, and for how long | So your firm knows when messages were not being checked | 90 days from when the pause began |
| Daily counts of what was caught and how often each feature was used, for each supported AI site how often SourceSecure started there, and the habit you are working on, from a fixed list (counts only, no text) | To show your firm's admins how SourceSecure is being used, for the firm or for one person | Until the firm's records are deleted. If you leave the team, the counts stay as the firm's and stop being linked to you |
| A summary of your week, once a week: counts, labels from fixed lists, how many chats fell in the morning, afternoon and evening, how many minutes a chat lasted, the short headlines of your last few notes (written by the AI model, not by you) and, on Guard 360 seats and above, a few short facts about your work from your vault (no message text, clock times or sites) | To write your private weekly note, which is kept on your computer. It is sent to OpenAI for that (see 4.2) | Not kept by us |
| A secret that proves your browser belongs to your seat, stored only as a one-way hash | To stop seats being copied | While the browser is connected |
| The firm profile, shared terms and firm playbook your admins write | To apply your firm's settings | Until the firm's records are deleted |
| Earlier versions of the firm profile, each with the email address of the admin who made the change | So a change can be undone | The last 100 changes. An admin's address is taken off them if they leave the team |
| An encrypted copy of your vault, your own list of names to catch and your list of things marked not sensitive (see 5.1) | So your vault is not lost when you remove the extension or change computer, and is the same on every computer you sign in on | Until you delete it from your vault, you are removed from your team, or your firm's records are deleted |
| On a Guard Enterprise firm, the team vault (see 5.3): short facts about the firm's work and the names of its clients, sent from Guard Enterprise seats' vaults or added by an admin; how many people's vaults agreed on each, counted under a one-way key for each person; what only one person's vault has sent so far, shown to nobody, while it waits for a second; and, for anything taken out, the email address of the person who took it out | To share how the firm works and who its clients are with everyone in the firm | Until it is taken out of the team vault, or the firm's records are deleted. What your vault sent stays if you leave the team |
| A request you make to your admin for a better seat: your email address, which seat, and when | So your admin can answer it | Until your admin answers it, or you leave the team |
| For a firm with Guard 360 seats or above, how many AI replies its people had checked each month, as one number for the whole firm | To keep the firm's monthly allowance | Until the firm's records are deleted |
| For each firm, each month: each request to the AI features as numbers only (which feature, which model, how many tokens and pages, what it cost), and the month's totals, including how many pages of documents its people had read. No text, and no person named | To keep the firm's monthly AI allowance and, on Guard Enterprise, its monthly pool of document pages, and to show the firm's admins what was used | Until the firm's records are deleted |
| Billing details (name, email, card, address), entered on Stripe's own page. We never see the card | To take payment | As Stripe and tax law require |
| The firm's billing name and address as given at checkout, the versions of our terms the firm agreed to, when, and the email address of the person who agreed | To make out the agreement to the firm, and to show what was agreed | Until the firm's records are deleted. The address of the person who agreed stays with this record even if they leave the team |
| Whether a payment has failed and is being retried | To tell your firm's admins, by email and on the team page | Until the payment goes through or the subscription ends |
| Which notices of a change we emailed your firm's admins, when, and to which addresses | To show that the notice our terms promise was given | Until the firm's records are deleted. An address is taken off them if that person leaves the team |
| After a firm's records are deleted: the firm's name, its Stripe identifiers, what it bought and when, which versions of our terms it had agreed to, and how many records were deleted. No person is named | The record of a sale that tax law expects, and proof that the deletion happened | 7 years after the deletion |
| A one-way key made from the buyer's email address, when a firm has had the free Guard 360 trial. It cannot be turned back into the address | So the free trial is given once to each buyer | 3 years |
| When each person's free Guard 360 trial began, filed under a one-way key made from their email address. It cannot be turned back into the address | So each person gets the trial once at a firm, even if they are removed and invited again | Until the firm's records are deleted |
| Server logs: request counts, timings and errors, with no message text | To run and secure the service | 30 days |
4. What happens to your messages and files
4.1 Detection happens on your computer. The extension reads what you type or attach on supported AI chat sites, finds sensitive details with models that are part of the extension, and offers a redacted version. None of this sends your text or files to us.
4.2 Optional AI features send a redacted extract to our server and on to OpenAI. Names, numbers and other details found are replaced with stand-ins such as [PERSON_1] first, and our server checks again for an email address, a phone number or a long number and refuses the request if it finds one. Redaction replaces what it finds; a name it does not recognise can pass through. We do not store the extract. We send it to OpenAI with OpenAI's storage option switched off, and OpenAI does not use it to train its models. OpenAI may keep its own abuse-monitoring logs of these requests for up to 30 days under its standard API terms, and does not use them to train its models. The features are:
- Prompt coaching (Guard Pro and above): parts of the conversation, yours and the AI's, and for Help me and Sharpen my prompt the whole conversation on the page and your unsent draft. Help me starts when a tip appears, and Sharpen my prompt when your pointer rests on its button for a moment, so the redacted conversation can be sent before you click. If you set Sharpen to Auto or Review, pressing Enter on a message also sends the redacted conversation and your draft to be sharpened. On Auto, the sharpened message is then sent to the AI site without a stop to look at it first; on Review, it waits for you.
- Weekly note (Guard Pro and above): once a week, the summary of your week described in section 3, to write your note. The note comes back to your computer.
- Reply marks (Guard 360 and above): your message, the AI's reply, a few earlier messages, and up to 25 facts from your vault (those closest in meaning to your message, picked on your computer) and your firm's profile.
- Clean-up questions (Guard 360 and above): at most once a week, up to 25 short facts from your vault, to ask which are worded too vaguely to check a reply against. Every answer is applied on your computer and nothing about it is sent.
- Chat and meeting summaries (Guard Enterprise): when an AI chat ends (you switch to another chat, close its tab, or leave it for 3 minutes), its last 40 messages, yours and the AI's, are sent once to be summarised for your vault. Until then, a copy with the names already replaced is held in your browser's memory, never on disk, so a chat can still be summarised after its tab is closed. At most 50 chats a day are sent per person. You can switch this off for yourself ("Remember my chats automatically"), for one chat ("Don't remember this chat"), or your firm can switch it off. A meeting transcript is sent as soon as you add the file.
- Ask SourceSecure (Guard Enterprise): your question, an overview of your vault, and the passages of your notes found for the question, sent only when you ask. A question asked on the team vault sends the team vault's facts and client names found for it instead of your own notes.
- Document reading (Guard Enterprise): when you add a document in your vault's Documents tab with "Teach my vault from it" ticked, its words, once, in parts, with names and client details replaced by stand-ins on your computer first, together with your note about it if you write one; and once for each batch, the short facts found in it with the facts your vault already holds. "Just keep it, send nothing" sends nothing, and your firm can switch document reading off. The document itself is never sent. Your firm shares a monthly number of pages, counted by our server from what it receives.
You can switch prompting tips off at any time. With it off, prompt coaching, the weekly note, reply marks, clean-up questions, chat and meeting summaries, and document reading all stop. Ask SourceSecure is separate: a question is sent only when you ask it, whether prompting tips are on or off, and stops when you or your firm switch your vault off. While SourceSecure is paused, none of these features sends anything, Ask SourceSecure included.
4.3 Real names in replies. When an AI's reply contains stand-ins we sent, the extension puts the real names back on your screen, using a list kept in your browser for up to 12 hours. The list is never sent to us.
4.4 Your voice. An AI site's own dictation and voice mode send your voice to the AI company before there is any text to check. So while SourceSecure is on (not paused or switched off), the extension holds the AI site's microphone on supported sites:
- Dictation. When you press the site's microphone or dictation button, SourceSecure listens instead, using your browser's built-in speech recognition set to work only on your computer. The words go into the message box and are checked like anything you type before you send them. No recording is made, kept or sent to us or to the AI site. Before first use, your browser may download its speech pack from its maker (for Chrome, Google); no recording is sent with that download. If your browser cannot recognise speech on your computer, SourceSecure does not listen at all, and says so.
- Voice mode. A spoken conversation with the AI site cannot be checked. It is held, and you may let it through once at a time; what you then say goes from your browser straight to the AI site, under that site's own terms, and SourceSecure receives none of it. Your firm's admin can turn this off, so voice mode stays held.
5. Your vault
5.1 Your vault is kept in your browser on your own computer, and an encrypted copy of it is kept on your SourceSecure account.
- On Guard 360 it keeps short facts about your work and firm, to check AI replies against. If you switch on Remember personal details, it also keeps lasting facts about your own life; while that switch is off, facts about your own life are not sent.
- On Guard Enterprise it also keeps notes, files, captures, people and company pages, summaries of your AI chats and meetings, and the documents you add with what was learned from them, encrypted.
Every computer you sign in on with the same address keeps the same vault: what you add or delete on one reaches the others within minutes. Your browser sends the copy every 10 minutes and about 30 seconds after a change, whatever your seat, and your own list of names to catch and your list of things marked not sensitive are copied with it, the same way. Before anything leaves your computer it is encrypted with a key made for you alone; our server keeps only the encrypted copy, never the key, and makes the key again from a secret held in Google's secret store when one of your signed-in browsers asks for it. So the copy in our database, or a backup of it, cannot be read on its own. We do not open it, share it with your firm, or use it to train AI models. If you are removed from your team, or your firm's records are deleted, the copy is deleted with your other records; the vault on your computer stays until you delete it. Only you can see it, and you can change or delete any of it; after a chat is summarised, Undo removes what was just added. Passages from it are sent, redacted and not stored by us, only for the features in 4.2, and, on a Guard Enterprise firm, the short work facts and client names described in 5.3 are shared with your firm's team vault. If you remove the extension or clear your browser data, it comes back from the copy when you sign in again. Delete everything, on your vault page, deletes it on every computer and from the copy. An export is not encrypted.
5.2 It may hold any fact you choose to keep, including a health fact you mention. On our servers it is held only as the encrypted copy in 5.1, apart from what is shared with a team vault under 5.3. A fact you keep may be sent, redacted, for the features in 4.2.
5.3 The team vault (Guard Enterprise). A firm with Guard Enterprise seats has one team vault: short facts about how the firm works and the names of its clients. It is held on our server, in Montreal, with the firm's other records, because the whole firm shares it.
- What is sent. About once a day, each Guard Enterprise seat's browser sends the short work facts its vault has learned about the firm, and the names of the clients it knows, at most 80 work facts, 80 facts from documents and 80 client names. Facts about you or your own life are never sent, and our server never opens your vault's copy to look for them. Nothing from a chat you set to "Don't remember this chat" is sent. A fact with an email address, a phone number or an account number in it is dropped. What is sent is not encrypted like your vault's copy: our server can read it, because it compares it and shares it with your firm.
- How something goes in. A fact or client name goes in once the vaults of two or more people in the firm have learned it. A work fact from a document a Guard Enterprise member adds goes in at once; a client name always waits for a second person. Until a second person's vault agrees, what only one person's vault sent is kept by our server, shown to nobody. The firm's admins can also add facts and rules themselves. The comparing is done by our server; nothing is sent to OpenAI for it. Nobody approves an entry first.
- Who sees it. The firm's team heads and Guard Enterprise members can see it, and any of them can take an entry out or put it back; each removal records the email address of whoever made it. Every seat in the firm uses it without seeing it: its facts help Sharpen and the reply check, and its client names are added to the list of names every seat in the firm catches and replaces before a message is sent.
- How long. An entry stays until someone takes it out or the firm's records are deleted. If you leave the team, what your vault sent stays in the team vault.
- Switching it off. A Guard Enterprise team head who switches the vault off for the firm stops every seat sending to the team vault and stops showing it.
6. Who else handles information
We use Google Cloud and Firebase (database and server functions in Montreal; website delivery and sign-in on Google's own network), OpenAI (redacted text for AI features, in the United States), Stripe (payments, in the United States and Canada), Resend (the emails we send, in the United States) and Cloudflare (our domain's addresses, and forwarding of email sent to support@sourcesecure.ca, on its global network). The current list is at sourcesecure.ca/subprocessors. We do not sell or share information with anyone else, except where the law requires it.
7. Where information is kept
Our database and the server functions that handle your information are in Montreal, Canada. Sign-in accounts are held by Google's Firebase Authentication, which is not tied to that region. Redacted text for AI features (OpenAI), the emails we send (Resend) and payments (Stripe) are processed in the United States, where they may be subject to US law. Email you send to support@sourcesecure.ca passes through Cloudflare's network on its way to us.
8. Your rights
You can ask to see the information we hold about you, have it corrected, or have it deleted, and (in Quebec) receive it in a structured format. Because your firm controls your account, we may pass your request to your firm's admin. Write to our privacy officer, Jacob Oliveira, at support@sourcesecure.ca; we reply within 30 days.
Your firm's admins can download everything we hold for the firm, as one file, from the team page, and can remove you from the team there, which deletes your records. The firm's owner can delete all of the firm's records there at any time. Two things stay if you are removed: if you agreed to our terms on your firm's behalf, the record that you did (your email address and the date) is kept with the firm's agreement; and if you had the free Guard 360 trial, when it began is kept under a one-way key made from your email address, so it is not given twice. Both are kept until the firm's records are deleted.
If you manage a team, we email you a monthly report; the link in it stops it in one click. If you are not satisfied, you can complain to the Office of the Privacy Commissioner of Canada (priv.gc.ca) or, in Quebec, the Commission d'accès à l'information.
9. Security
We encrypt information in transit and at rest, keep no downloadable server credentials, require two-step verification for our administrative accounts, and keep the database protected against deletion with 7-day recovery. The copy of your vault is encrypted before it leaves your computer, and the secret its keys are made from is held apart from the database. If a breach affects your information, we will tell your firm within 72 hours and help it notify you where required, as our incident and breach response policy at sourcesecure.ca/incident-response sets out.
10. Cookies
sourcesecure.ca uses no advertising or tracking cookies. Signing in stores a session in your browser so you stay signed in.
11. Children
SourceSecure is for businesses and is not meant for anyone under 18.
12. Changes
If we change this policy in a way that affects what is collected or who receives it, we will email firms' admins before the change takes effect. The date at the top shows the last change.
This is version privacy-2026-11-11, also kept at its own address. Every version of this document.
The detail, feature by feature
The policy above is the whole policy. What follows describes, part by part, exactly what SourceSecure sends and keeps, in more detail than a policy needs, for whoever has to check it. If anything here ever reads differently from the policy above, the policy above applies and we will correct this part.
What stays on your computer
Every message you type on a supported chat site is checked in the browser. The model that recognises names, addresses and company names runs on your machine, inside the extension. It is a file on your disk, not a service we operate.
Attachments are read the same way. A PDF, Word file, spreadsheet, PowerPoint or scan is opened, read and redacted in the browser. The file is never sent to us, and neither is the redacted copy. The redacted copy goes only where you attach it: to the AI site, in place of the original.
Your own word lists and the list of things you have personally chosen to ignore are held in your browser's storage, and an encrypted copy of them is kept on your SourceSecure account with your vault, as described below. Your messages reach us only as described under the AI features below: redacted first, and not stored by us.
On a Guard 360 seat, your vault is held there too: short facts about your work and your firm, learned from your messages or saved by you, which AI replies are checked against, and facts about your own life too if you switch on Remember personal details. On our servers it is held only as the encrypted copy described below. Its facts are sent, redacted, with each reply check, as described below. You can see, change and delete every fact on its own page, reached from the SourceSecure icon.
We never store your prompts on our servers. On a Guard Enterprise seat, your vault also keeps your notes, files and short summaries of your AI chats and meetings. Your vault is encrypted on your computer, and an encrypted copy is kept on your SourceSecure account, so it comes back if you remove and add the extension again and is the same on every computer you sign in on. Your browser sends the copy every 10 minutes and about 30 seconds after a change. Our server keeps only the encrypted copy and never stores its key: it makes the key again from a secret held in Google's secret store, only for your own signed-in browsers. We never open the copy, share it with your firm or use it to train AI models. On Guard Enterprise, the short work facts and client names your vault learns are also shared with your firm's team vault, as described below.
When you ask your vault to learn from a document, a copy with names and client details replaced by stand-ins is read once by our server, which keeps nothing. The document itself is never sent to be read; it and what was learned are kept in your vault.
What we hold about your account
Your email address, which is how you sign in and how we know which firm you belong to.
Your firm's name, how many seats of each kind it has bought, which kind of seat each person has, and whether the subscription is in good standing.
Who the firm's owner is: whoever bought, until they hand it to another admin, and each time it changed hands. The owner's address is also the one on the firm's customer record at Stripe.
The billing name and address the buyer gave on Stripe's checkout page, which is the address the firm's agreement is made out to. Which version of each document the firm agreed to, when, and the address of the person who ticked the box; an admin can see these on the Billing tab, each linked to the exact text.
Whether your firm chose the two weeks of Guard 360 at checkout, and when each person's two weeks end. A one way key for the buyer's address is kept for three years so the trial is only given once; it cannot be turned back into the address.
If somebody asks their admin for a better seat (to keep Guard 360 after their two weeks, or for Guard Pro or Guard 360 from the extension’s setup), their email address, which seat, and when they asked, which their admin sees. They are told so before they ask. It is removed when the admin gives them the seat, says not now, or removes them.
For a firm with Guard 360 seats, how many replies its people have had checked this month, as one number for the whole firm, so the firm's monthly allowance can be kept.
Which browsers you have connected, when each one was connected, and when it last checked in. Each browser holds a secret; we hold only a one way hash of it, never the secret itself.
Identifiers from Stripe for your firm's customer and subscription records, and, while a payment that failed is being retried, that it is. We never see or hold card details.
For each notice of a change we email your firm's admins, as our terms promise: which notice, when, and to which addresses it went.
What the daily counts contain
Once a day each browser sends a small report so your admin's dashboard can show whether the firm is using SourceSecure. With the browser's id and the secret that proves it, it contains dates and counts, and nothing else. The counts are: how many items were caught, and of which kinds; how many times a message was held for review, and how many times it was then sent redacted or sent as typed; how many attached files were checked, flagged, and replaced with a redacted copy; how often Set up this chat, Help me and Sharpen my prompt were opened and used; how many tips were shown; how many prompts were scored and the scores added together, before and after; how many weekly notes were written and read, and plans kept; and, on Guard 360 seats, how many replies were checked, how many marks were shown, opened, asked about and dismissed, and how many replies drew each kind of mark, by its name from a fixed list such as “assumed” or “decided for you”. A file is counted, never described: its name and its contents stay on your computer.
With coaching, on Guard Pro, Guard 360 and Guard Enterprise seats, the report also names the habit the person is working on, from a fixed list, as a count of one.
The report also says, for each supported site, how many times SourceSecure started there and how many times it could not find the message box. This is the one thing in the report that names anything outside your own browser, and what it names is one of the site names already printed on this page, never a web address, a page title or anything you were looking at. It is there because a chat site that redesigns its page stops SourceSecure working without any sign that anything is wrong, and these two numbers are the only way we would ever know.
There is no field in that report through which a sentence could travel. The code that reads it accepts a date and numbers, and discards everything else; a test walks every field and fails if any of them could carry text.
Each browser's counts are stored under your firm, with the browser's id and a key made from the email address of the person it belongs to, so that somebody with several browsers counts as one person. Your admin sees these for the whole firm, or for one person. For each person they also see every catch from the last 90 days: when, on which AI site, what kind of detail, and whether it was redacted, sent as typed or marked not sensitive, with the detail itself masked on the person's computer before it is sent (for example •••• 4242 or Rob••• Ch••). The mask keeps part of it so the record can be recognised: the last four digits of an account number, the last two of a SIN or phone number, the first half of each word of a name, an address or other words, the start of an email's name and of its domain with its ending (jane••••@smit••••.ca), and the first part of a postal or ZIP code (N3C •••). The full detail and the rest of the message never reach us. They also see when protection was paused or switched off on each person's browser, or its checking set to manual (only when the person asks) or off, and for how long. Catches are deleted 90 days after they are received, and a pause 90 days after it began. Both are deleted at once when somebody leaves the team.
The word lists your firm sets
An admin can add words that SourceSecure should always redact, such as the firm's own name, a client company, a fund or a project codename. They can also add words it should stop flagging, such as an internal reference that looks like an account number.
These lists are typed by your admin and stored by us, because every browser in the firm has to receive them. If your admin types a client's name into the always redact list, we hold that name. The lists, the firm's profile and the firm's playbook hold a client's name only if somebody at your firm typed it there. On Guard Enterprise the team vault also holds client names, which people's vaults learned and sent, as described below; and a person's own vault can hold names, which reach us only inside its encrypted copy. A name the detector fails to recognise can also pass through, unstored by us, inside the redacted text of an AI feature; the terms say plainly that it will sometimes miss one.
They travel to each browser with the daily licence check. An admin can remove any entry at any time, and the removal travels the same way.
Your firm's profile and team vault
An admin of a firm with Guard 360 seats can keep a profile of the firm: short facts about it, such as which software it uses, and rules for how AI is used there. Like the word lists, these are typed by your admin and stored by us, because every Guard 360 seat in the firm receives them with the daily licence check. They are included, redacted, when a reply is checked.
On Guard Enterprise the profile is the firm's team vault, and it also fills itself. About once a day each Guard Enterprise seat's browser sends us the short work facts and client names its vault has learned about the firm. Each time, at most 80 work facts, 80 facts from documents and 80 client names are sent. Facts about the person's own life are never sent, nothing from a chat set to Don't remember this chat is sent, our server never opens their vault's copy to look for them, and a line with an email address, a phone number or an account number in it is dropped. These facts and names are not encrypted like a vault's copy: our server reads them, to compare them and share them with the firm. A fact or client name goes in once two or more people's vaults have learned it. A work fact from a document a Guard Enterprise member adds goes in at once; a client name always waits for a second person. Our server does the comparing; nothing is sent to OpenAI for it. We keep a one way key for each person who sent a fact, only so that one person is not counted twice. Until a second person's vault agrees, what only one person's vault sent is kept by our server, shown to nobody.
Nobody approves an entry first. The firm's team heads and Guard Enterprise members can see the team vault, and any of them can take something out or put it back; each removal records the email address of whoever made it. Every seat in the firm uses it without seeing it: its facts help Sharpen and the reply check, and its client names join the names every seat catches before a message is sent. An entry stays until someone takes it out or the firm's records are deleted, and what a person's vault sent stays if they leave. A team head who switches the vault off for the firm stops it.
Prompt coaching, on Guard Pro, Guard 360 and Guard Enterprise seats
Guard Pro, Guard 360 and Guard Enterprise seats can suggest how to ask a better question. Doing that needs a model to read the conversation, and that model is not on your machine.
Before anything is sent, the same redaction that protects a message protects this: the names, account numbers, SINs, addresses and the rest that it finds are replaced with stand ins such as [PERSON_1]. The redacted text is then sent to our server, which passes it to OpenAI, which returns a suggestion.
Our server checks the text again before it goes out. If it still finds something that looks like an email address, a phone number or a long number, it refuses to send rather than cleaning it quietly, because a leak getting through means the extension has a fault worth knowing about.
We store nothing from a coaching request: it is not written to our database or our logs. It is sent to OpenAI with OpenAI's storage option switched off (store: false). OpenAI may keep its own abuse-monitoring logs of these requests for up to 30 days under its standard API terms, and does not use them to train its models.
Tips look at the messages of the conversation you are having now, a few at a time, both your messages and the AI's replies. Help me, Sharpen my prompt and Set up this chat look further. Help me and Sharpen my prompt read the whole conversation on the page, including messages from before you opened it and anything typed in the message box but not yet sent, and any answers you give to their follow up questions. Set up this chat sends the answers to its questions when you press Write. On a Guard Enterprise seat, any vault notes you tick in Help me, Sharpen my prompt or Set up this chat are sent with them. All of it is redacted the same way before it leaves.
So that Help me is ready the moment you click it, it starts reading as soon as a tip appears on screen, before you have clicked anything. That means the redacted conversation is sent whenever a tip is shown, whether or not you go on to use Help me. Sharpen my prompt starts early in the same way: resting your pointer on its button for a moment, with something typed, starts preparing it, and that sends the redacted conversation then, before you click. We kept this because it makes the button answer faster. It is limited to 60 a person a day. With Sharpen on Manual, what comes back is placed in the message box for you to read and edit; it is never sent on your behalf. With Sharpen on Auto or Review, pressing Enter on a message sends the redacted conversation and your draft to be sharpened. On Review, the sharpened message waits on a card for you; on Auto, it is sent to the AI site without a stop to look at it first. Short messages and small talk are sent as typed, and a rewrite that is no better is not used.
Once a week coaching also writes each person a short note about how their week with AI went. For that, a summary of their week is sent through our server to OpenAI, as coaching is: how many chats, how many messages each took, how that compares with their own earlier weeks, which of a fixed list of habits the coach noticed, and how many messages chats of each kind of work took, from a fixed list (client emails, summaries, research, analysis, drafting) the coach tags as it reads, never the subject. It contains no text anyone wrote, no names of sites and no clock times. On a Guard 360 seat it also carries up to eight short facts from the person's vault about their work, such as “Most of your clients are retirees”, learned from their redacted messages and checked again for client details, so the note's example can be set in their line of work. It does say how many chats fell in the morning, the afternoon and the evening, on how many days, and how long one notable chat lasted, and it carries the short headlines of the person's last few notes, which the coaching model wrote, so a new note does not repeat them. The note comes back and is kept on their computer only. Their employer never sees it.
Each person's browser also tells us which habit, from that fixed list, they are working on, so their firm can see which habits its people are building. For the whole firm, the admin sees only totals: nothing at all until at least five people have picked a habit, and a habit fewer than three people share is counted under other habits. An admin who picks one person sees that person's habit and counts. Nobody's note or plan, in their own words, is ever shown to the firm.
While an AI chat site is open and in view, a browser with coaching on also sends a short wake-up call to the coaching service every few minutes, so that a tip or Help me does not have to wait for it to start. The call carries nothing: no text, no id for the browser or the person, and not which site is open. Like any request, it does tell the service that some browser was using an AI site at that moment, from the network address it came from.
Coaching can be switched off by any person for their own browser, at any time, from the SourceSecure icon in the browser's toolbar. With it off, coaching sends nothing: no tip, Help me, Sharpen my prompt, Set up this chat or weekly note is sent, reply marks stop with it, and the wake-up call stops too. The daily counts, the masked catches and your pauses still go to your firm. On a Guard Enterprise seat a question you put to Ask SourceSecure is separate: it is sent only when you ask it, redacted first, whether prompting tips are on or off. It is described below. Pausing SourceSecure stops every one of these, Ask included.
Reply marks, on Guard 360 and Guard Enterprise seats
Everything here also applies during a two-week Guard 360 trial, if your firm chose one: for those two weeks your seat works as Guard 360, then goes back to your own.
A Guard 360 seat checks each AI reply for the places worth a second look, such as a decision made for you, an assumption, or something that contradicts what your firm has told it, and underlines them faintly. That also needs a model that is not on your machine.
When a reply finishes, your message and the reply are redacted in your browser in the same way as coaching, with the last few messages of the chat before them (at most four, cut short) and up to six short sentences of your own from earlier in the chat that set a condition, such as a deadline or a limit, so the check can tell what you already said from what the reply assumed. All of it is sent through our server to OpenAI, together with the facts from your vault and your firm's profile, redacted the same way. Our server checks all of it again before it goes, as it does for coaching. What comes back is where to underline and a short note for each, which are shown only on your screen.
When your message says something about your work or your firm, the same redacted message is sent once more to read short facts from it, together with the facts already in your vault, redacted the same way, so that a fact which has changed can replace the old one. The facts come back and are kept in your browser's vault, with the old wording kept there as history. They leave your computer only inside the encrypted copy of your vault and, on Guard Enterprise, as the work facts shared with your firm's team vault. The model is told never to keep anything about your private life, your health, your family, your own money or your job search, and never anything about a client or another named person.
Remember personal details is off unless you switch it on in the popup. While it is on, lasting facts about your own life, such as an allergy, your family or what you like, may be kept in your vault as well. They stay in your vault and are never offered to your firm, but like your other vault facts they are sent, redacted, with each reply check. Nothing about another person or a client is kept either way. Switching it off stops them being learned and sent; those already kept stay in your vault until you delete them.
Nothing from a reply check or a fact read is stored by us, and both are sent to OpenAI with its storage option off, as coaching is. Our logs record how many marks or facts came back and how long it took, never any words.
A firm's Guard 360 seats share an allowance of checked replies each month. Past it, every reply is still checked, but only the most important marks are shown until the 1st, and a person can have up to 20 replies a day checked in full. The monthly total is kept for the firm as one number, as described above.
Reply marks have their own switch, Mark AI replies, under the SourceSecure icon, and switching prompting tips off stops them too. With prompting tips off, no reply check and no fact read is sent. The two things a Guard Enterprise seat can still send when you press for them are described under prompt coaching above and the vault below.
Your vault, on Guard Enterprise seats
A Guard Enterprise seat comes with the full vault: your notes, the documents you add with what was learned from them, what you capture from web pages, and short summaries, decisions and deadlines from your AI chats and meetings. It is kept in your browser, encrypted, and an encrypted copy is kept on your SourceSecure account so every computer you sign in on has the same vault, and it comes back if the extension is removed. Our server keeps only the encrypted copy and never stores its key; it makes the key again from a secret in Google's secret store, only for your own signed-in browsers, and never opens the copy. The work facts and client names your vault learns are also shared with your firm's team vault, as described above. Pieces of it are sent, redacted, only for the things described below, and are not stored by us.
Searching it, by words and by meaning, happens on your computer, with a small model that is a file inside the extension. What you search for is never sent anywhere.
When an AI chat ends (you switch to another chat, close its tab, or leave it for 3 minutes), it is summarised for your vault without asking.
While the chat is open, your browser keeps a copy of it with names and details already replaced by stand-ins, in memory only and never on disk, so a chat can still be summarised after its tab is closed; the copy is dropped once it is read. Before it is sent, your computer checks again that no real name or detail the chat used is left in it, and does not send it if one is.
The last 40 messages, yours and the AI's, are sent once through our server to OpenAI, which writes a short summary and lists its decisions and deadlines. At most 50 chats a day are sent per person. The real names are put back on your computer, and the result is filed in your vault. A short note then says how many things were added, with Undo to remove them.
Neither our server nor our logs keep any of the chat or the summary; the logs record how many decisions and deadlines came back. Nothing already in your vault is sent to compare it with.
Nothing is sent:
- if you switch off Remember my chats automatically on your vault page
- for a chat where you chose Don't remember this chat, in the SourceSecure menu on that page
- if your firm switches the vault off
- with prompting tips off
- while SourceSecure is paused
A meeting transcript you add to your vault, from Zoom, Teams or Google Meet, is summarised the same way, and this starts as soon as the file is added: there is no second press, and the place where you add the file says so. The transcript is sent in parts, with every speaker and each name and detail found replaced by a stand-in. The transcript with its real names, and the summary, are kept only in your vault. With prompting tips off, or while SourceSecure is paused, the transcript is kept in your vault and nothing is sent. A document you add in the Documents tab with Teach my vault from it ticked is sent to our server to be read, once, in parts, with names and client details replaced by stand-ins on your computer first, as the policy above says; the server keeps nothing. With Just keep it, send nothing, it is kept in your vault and nothing is sent.
Ask SourceSecure answers questions about your own vault. When you ask, your question is sent with an overview of the vault (how much is in it, the people and companies as stand-ins, recent titles, open items and headings), the passages your computer found for the question, and your last three questions and answers. To answer, it may ask your computer to look further, up to four more times, and what that finds is sent too. All of it is redacted first and checked once more on your computer, and the request is not sent if a name the vault knows or a number is still in it. You can see what was sent under each answer. A question asked on the team vault sends the team vault's facts and client names found for it instead. A question is sent only when you ask it. That does not depend on the prompting tips switch, because asking is a button of its own; it does stop while SourceSecure is paused or switched off, when the box says so and nothing is sent, and when you or your firm switch the vault off.
Your firm's shared playbook (its templates, rules and approved wording) is held on our server, like the team vault, because it is shared. Only your firm's team heads can change the playbook, a change made in the extension is checked there for client details, names included, and every change is checked by our server, which refuses anything that looks like an email address, a phone number or a long number. A change made on the team page is not checked for names, and the page says so. Its rules are sent with each reply check, like your firm's profile. Of your own vault, our server holds only the encrypted copy.
You can export your vault as a folder of notes, which is not encrypted, and delete any of it, or all of it, on its page. Your firm can switch the vault off; what you already kept stays until you delete it. The vault's key is kept by the browser and is not tied to a password, so anybody who can use your browser profile can open your vault.
On a Guard Enterprise seat you can also add to your vault from any web page, from the right-click menu: a part of the screen, the text you selected, or the page's readable text. The extension reaches that page only when you pick one of these, and only that page. A part of the screen is read for text on your computer, by the same reader that reads attached scans. You see what will be kept, and change it, before you press Save; the text, the page's title and address (without anything after a ? or #), and the picture if you keep it are stored encrypted in your vault, like everything else in it. Nothing you capture is sent to us when you save it. Asking about a selection from the same menu sends that selection as a question to Ask SourceSecure, as described above.
Where it is held
Your firm's records and the daily counts are held in Google Cloud Firestore in the northamerica-northeast1 region, which is Montreal, Canada. The functions that serve this site and the API run in the same region.
Some things are held elsewhere, by the companies named below. Sign in is run by Google's Firebase Authentication, which keeps each person's email address and sign in history in its own systems rather than in Montreal. Stripe holds billing details and the firm's name and its owner's address, in the United States and Canada. Resend handles the email addresses we send to, in the United States. Email you send to support@sourcesecure.ca is forwarded to our mailbox by Cloudflare, which also runs the site's addresses. Prompt coaching and the weekly note, reply marks and, on Guard Enterprise, Ask SourceSecure, chat and meeting summaries and document reading send redacted text to OpenAI, in the United States. Our own server logs record what happened, never what anybody wrote, and are kept by Google Cloud.
Nobody else
The companies in section 6 above, listed in full on the subprocessors page, are the whole list. OpenAI receives text only after it has been redacted on the computer it was typed on, only for people on Guard Pro, Guard 360 or Guard Enterprise seats, and with no account details: nothing that says which person or firm it came from. We do not use advertising networks, analytics services or trackers, and this site sets no cookies of its own. It loads its own fonts from its own servers rather than from Google, because a page selling the promise that nothing calls out should not quietly call out.
How long we keep it
Account records are kept while the firm has an account with us. When an admin removes somebody from the team, their membership, their connected browsers, their catches and pauses, their sign in account and the counters that limit how much they can use are deleted, the firm's daily counts stop pointing at them, and their key comes off any suggestion for the firm's profile. The counts themselves stay, as the firm's.
Their address also comes off the firm's own records that used to carry it: who made each saved version of the firm's profile, who sent an invitation still waiting, the receipts for the reports and notices they were emailed, and the list of the firm's owners. If they were the owner, the firm passes to the admin who has been on the team longest, and Stripe's customer record is given that address instead. One record keeps their address on purpose: if they ticked the box agreeing to our terms for the firm, the record of who agreed and when stays with the firm's agreement, because it is the firm's proof of its own contract. It goes when the firm's records do. Stripe keeps its own records of a buyer, such as invoices, as the law requires of it.
The firm's profile and the suggestions for it are kept until the admin removes them, or until the account is closed.
Daily counts are kept so a dashboard can show a trend. They contain no text. They are kept for as long as the firm is a customer; when somebody leaves the team, their counts stay as the firm's and stop pointing at them.
When a firm's subscription ends, everything we hold for it is deleted 30 days later, automatically: its people, invitations, connected browsers, catches, pauses, counts, profile and its history, word lists, policies and playbook, and its people's sign in accounts. What stays is what the law makes us keep about a sale: the firm's name, its Stripe identifiers, what it bought and when, which versions of our terms it had agreed to, and a note of the deletion itself, with how many records went. It names no person, and it is itself deleted seven years later. The admin's team page says the date beforehand. A payment that has failed and is still being retried is not an ending, and deletes nothing; if every retry fails, the subscription ends and the 30 days start then.
A firm does not have to wait for that, or write to us. Its owner can delete it at any time from the Billing tab of the team page, by typing the firm's name: the subscription is cancelled at once, with no refund for the time left, everything above is deleted straight away, and the firm's admins are emailed that it was done. Any admin can first download everything we hold for the firm, as one file, from the same tab.
If you want to see, correct or delete what we hold about you yourself, write to support@sourcesecure.ca. Your firm controls your account, so we will normally pass the request to your firm's admin, and we answer within 30 days either way.
Getting in touch
Email support@sourcesecure.ca. SourceSecure is operated by JOLIVE Labs Inc.