Privacy Policy

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 documents, your own word lists and the list of things you have personally chosen to ignore are held in your browser's storage. We cannot read them. 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. It is never stored on our servers. Its facts are sent, redacted, with each reply check, as described below, and a fact is held by us only if you choose to suggest it to your firm. 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, the Living Vault keeps your notes and short summaries of the chats and meetings you save, encrypted on your own computer.

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 R••• C•••). The mask keeps a little 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 letter of each word of a name. 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, 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 are the places a client's name is stored by us, and it gets there because somebody at your firm typed it there. 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, on Guard 360 seats

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.

A person on a Guard 360 seat can suggest a fact from their own vault for the profile. Only when they press Suggest to your firm beside it, and only the words of that one fact, are sent to us. Our server refuses a suggestion that looks like an email address, a phone number or a long number.

A suggestion is held for the admin, who sees its words and how many people suggested it, never who. With it we keep a one way key for each person who suggested it, only so that one person is not counted twice. It stays until the admin approves or dismisses it, or until the list of suggestions is full and it is the oldest that only one person made. When somebody is removed from the team, their key comes off every suggestion, and a suggestion nobody else made goes with 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), and OpenAI does not train on data sent through its business API. OpenAI may keep its own abuse-monitoring logs of these requests for up to 30 days under its standard API terms; we have asked OpenAI for zero data retention, and this sentence will change when it is in place.

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. What comes back is placed in the message box for you to read and edit; it is never sent on your behalf.

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: 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. Two things on a Guard Enterprise seat are separate, and each is sent only when you press it, redacted first: a question you put to Ask SourceSecure, which is sent whether prompting tips are on or off, and Write background in Add notes, which is sent only while prompting tips are on. Both are 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 and nowhere else, with the old wording kept there as history. 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. Save to vault does the same with text you select and choose to save.

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 browser 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, on your computer, 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 Living Vault below.

The Living Vault, on Guard Enterprise seats

A Guard Enterprise seat comes with the Living Vault: your notes, the files and folders you add, replies you choose to save, and short summaries, decisions and deadlines from chats you choose to save. It is kept in your browser, locked with a key the browser will not let anything read out, ours included. The vault is never uploaded to us and never synced anywhere. 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 a chat ends you are asked whether to save it. If you chose Always save in your vault's settings, it is saved, and sent, without asking; with Never save, with prompting tips off, or while SourceSecure is paused, nothing is sent. Saving sends the last 40 messages of that chat, yours and the AI's, once, redacted in your browser the same way as coaching, through our server to OpenAI, which writes a short summary and lists its decisions and deadlines. The real names are put back on your computer and the result is kept in your vault. 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.

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. Other files you add are read on your computer and are not sent.

Add notes, on an AI chat page, finds notes in your vault that bear on what you are writing. Finding them, searching and ticking them happen on your computer. When you press Write background, the notes you ticked (eight at most, up to 1,500 characters of each), the message you are writing and your own last four messages in that chat are sent, redacted, through our server to OpenAI, which writes a few sentences of background for you to read, change and add. The AI's replies are not sent, and none of it is stored by us. Add the notes as they are puts the ticked notes into the message box with names as stand-ins and sends nothing to us. Write background is sent only when you press it, and only while prompting tips are on and SourceSecure is not paused. Nothing goes into your message that you did not tick or approve.

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 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 Living Vault off.

Your firm's shared playbook (its templates, rules and approved wording) is the one part held on our server, because it is shared. Only your firm's team heads can change it, 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.

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 Living 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, reply marks and, on Guard Enterprise, Ask SourceSecure, chat and meeting summaries and Write background 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.