Privacy Policy
Closehold is built on a simple principle: we protect your data by never holding it. Closehold detects sensitive information the moment before it would cross into an AI tool and replaces it with reversible placeholders — and it does this on your own device, not on our servers. This policy explains, in plain terms, what the Closehold browser extension, desktop application, and website do with data, and what they deliberately never do.
This policy covers the Closehold browser extension, the Closehold desktop application it pairs with, and closehold.ai. It applies to everyone who uses Closehold, including pilot and evaluation participants.
The short version
- The extension processes the prompts and files you send to AI tools
only on your own machine (via a local companion app at
127.0.0.1). Nothing is sent to Closehold to be scanned. - The extension itself has no account, no analytics SDK, no advertising, no cross-site tracking, and collects no content. The desktop application keeps a diagnostic record on your computer, described below.
- If your computer is part of an organization’s Closehold account, the desktop application shares payload-free records with Closehold, on by default: a dated record of each protection and of each AI reply Closehold checked, which your organization’s administrators see per person. They never contain what you typed, your files, file names, or a sensitive value. These records are personal data, and we treat them that way.
- Service health diagnostics, linked to your license key, are not shared in this release: the diagnostic record stays on your computer and leaves it only in a support report you save and send yourself. A later release will share them, on by default for every enrolled organization, with a switch for your administrator; what they will contain is described below.
- If you choose Secure Send for an email, the values you flagged (and any file you share) are encrypted on your computer and held on our share service until the link expires or you revoke it. The key travels only in your link, so we cannot read them.
- We never store raw sensitive values — anywhere. Where a record of a data crossing is kept, it holds only classifications, counts, and salted one-way hashes that describe the shape and flow of data, never the value.
- We never sell your data, and never use it for advertising or to determine creditworthiness.
What the browser extension does
On the AI sites it supports (such as ChatGPT, Claude, Gemini, and Microsoft Copilot), the extension reads the prompt you are about to send and any files you attach, so it can protect confidential material in place before the prompt leaves your browser. When the AI’s reply comes back, the extension restores your real values in that reply so the conversation reads normally. The AI service never receives your original values; you never have to look at placeholders.
The extension is a thin bridge. It contains none of the detection, classification, or transformation logic itself — it hands content to the Closehold desktop application running locally on your machine, which does that work and hands back a protected version. The extension activates only on the AI surfaces it supports and does not read your general browsing.
The email advisory option
Closehold offers an opt-in email advisory for Gmail and Outlook on the web. It is off by default; you turn it on from the extension. When on, it reads the email draft you are writing — on your own device, via the same local application — and flags content that should probably not leave in an email (credentials, financial details, government identifiers, health information, confidential business content) before you press send. It is advisory only: it never rewrites your email on its own, and it never blocks one — when a draft is flagged, Closehold asks once before the send proceeds, and one click sends it unchanged. It never transmits your draft anywhere except to the Closehold application on your own machine, unless you choose Secure Send, described next. Ordinary email content — names, addresses, phone numbers — is deliberately not flagged. Typing and drafts are never recorded; when you make a send decision over a flagged draft, Closehold keeps only a payload-free record (categories and counts — never your text, never an address). When the option is off, the extension does not read your mail at all.
Secure Send
Secure Send is an optional part of the email advisory, in beta and off by
default. When you choose it for an email, the values you flagged — and any
file you share with it — are encrypted on your computer before they
leave it, and the encrypted copy is stored on Closehold’s share service
(share.closehold.ai, on Google Cloud). Your email carries a link in their
place. The key that opens the copy travels only inside that link, in the
part of the address a browser never sends to a server, so Closehold cannot read what it
stores: not the values, not the file, not the file’s name. The recipient’s
browser decrypts it.
Alongside the encrypted copy, the share service keeps when the share was made and when it expires, whether it opens only once, how many times it has been opened and when it was first opened, whether it was revoked, a stored file’s size, and a one-way hash of the computer that made it, so that only that computer can revoke it. Each request’s network (IP) address is used, in memory, to limit how fast the service can be called, and our hosting provider’s request logs record it.
A share stops opening when it expires (you choose 1, 7 or 30 days), when you revoke it, or, for a one-time link, after it is first opened; its record is then deleted. A stored file is deleted with its share, and in any case no later than 31 days after it was uploaded.
What we collect — and what we don’t
Content you send to AI tools is transmitted only to your own device
(127.0.0.1) for local processing. It is not transmitted to
Closehold or to any third party by the extension.
The extension uses your browser’s local storage to hold your on/off preferences, the local pairing state that links it to the desktop application, and short payload-free logs of its recent notices and protection outcomes (for example, that a send was protected or refused, and why), which help diagnose a problem. None of it leaves your browser except to the desktop application on your own computer. It contains no browsing history, no page content, and no prompts. There is no account, login, cookie, analytics SDK, advertising identifier, or cross-site tracker in the extension.
Where processing happens
Core detection runs locally — on your own machine, or, for organizations that choose it, on infrastructure the organization itself operates. In the organization-hosted mode, prompts are processed by a server the organization controls inside its own network (device → organization server → nothing); they are never sent to Closehold and never sent to a public cloud. Closehold’s local AI models run entirely on the device or the organization’s server.
What we never store
Closehold does not store raw sensitive values: not in logs, not in audit events, not in data-flow records, not in any default storage. Where we need to record that a crossing happened — for your own audit trail — we keep only classifications (for example, “a person’s name was protected”), counts, and salted, one-way hashes and token references that describe the shape and flow of data. The original value cannot be recovered from these records. In Closehold’s own terms, the raw value is never stored.
Reversible protection and encryption
So that your real values can be restored into an AI’s reply, Closehold keeps a local mapping between a placeholder and the value it stands for. This mapping lives only on your device. When it is persisted, it is encrypted at rest with AES-GCM, with the key held in your operating system’s keychain. It is never synced to us, and it is removed when you clear Closehold’s data or uninstall the application.
AI providers
When you use an AI tool through Closehold, that provider receives the protected version of your prompt — with sensitive values already replaced — not your original content. Closehold does not send AI providers anything beyond the protected prompt you were already sending them. How a provider handles the protected prompt it receives is governed by that provider’s own privacy policy; the purpose of Closehold is to ensure the provider never sees the raw values in the first place.
The diagnostic record on your computer
The desktop application keeps a diagnostic record on your own computer of how each protection went: how long it took and which step took the time, whether it was protected, refused, cancelled or failed and why, how many of each kind of sensitive information it found, and the shape of a file (its type, a size range, and a page or row count). It never holds what you typed, the content of a file, a file name or path, a page title or address, or a sensitive value. It is kept for 45 days and then deleted. You can preview it and save a copy from Settings to send to support; nothing in it is sent anywhere by that step unless you send the saved copy yourself. Anyone who administers your computer can also read files on it, including this record.
When your computer is part of an organization
When a computer is enrolled in an organization’s Closehold account (for example, by your employer), protection analytics are sent to Closehold’s servers, on by default for every enrolled organization. A second kind of record, Service health, is described below; it is not shared in this release. Both are payload-free: they never contain your prompts, your files, file names, or a sensitive value. They are not anonymous, though — they are linked to you, so we treat them as personal data.
Protection analytics, for your organization. A record, with the date and time to the second, of each protection and of each AI reply Closehold checked.
- For each protection: the kind of information that was protected (for
example, “a person’s name”), how sensitive it was, a risk level for
what still crossed, the AI service it was headed to, the kind of crossing (for example,
a prompt or a file upload) and how trusted that destination is, what Closehold did, where the content came from (typed, pasted, or a file,
and the file’s format — never its name), the placeholder that replaced the
value (such as
[PERSON_1]), the protection policy and version in force, and salted one-way hashes that let repeated crossings be counted without revealing the value. - For each AI reply Closehold checked on a supported site — including a reply where nothing needed protecting: the AI service, the AI model when the site names it, a reply-quality label (one of: clean, invalid format, not in the format asked for, or repeating itself), the type of task your prompt asked for (one of: code, summarize, translate, rewrite, extract, classify, question, or chat), and, if the reply contained sensitive information, its kind and what Closehold did. Never the reply and never the prompt.
- From Closehold Clear and Verify: counts of reviews, decisions and answers checked, and how often a document already protected was sent again; and, for each document Clear made a safe version of, when it was made, its format, the Clear profile used, and how many items of each kind were redacted, rewritten or kept.
Your organization’s administrators can see these per person in the Closehold admin console — which kinds of information were protected, where, when and how often, and what kinds of task were asked of which AI service — never the information itself. Closehold staff can also read them, only to support your organization. Enrolling the computer is what turns this on.
Enrollment, license checks and your organization’s rules. To enroll a computer and keep its license current, the desktop application sends Closehold’s servers the computer’s name, a random id for this installation, a device key, and the Closehold version, and then checks the license from time to time. An enrolled computer also downloads its organization’s protection rules. None of these requests carries anything you typed or opened.
Service health, for Closehold. Not shared in this release. In this release the desktop application sends no Service health: the diagnostic record above stays on your computer, and leaves it only in a support report you save and send yourself. A later release will send Service health, on by default for every enrolled organization. What it will send: diagnostics drawn from the record above, linked to your license key, so we can find and fix failures and slow uploads: how long protection took (in ranges) and which step took longest; whether it was protected, refused, cancelled or failed, and a reason code; the shape of a file (its format, a size range, and a page or row range); which AI website or desktop app was involved, and how the file reached Closehold (for example, a paste, a drag-and-drop or the file picker) and what was sent in its place (the same format, a converted file or a text copy); which internal component or detector failed or was not running, and for how long; whether the diagnostic record is switched on; how busy the computer was (how many scans were running, and processor and memory load in ranges); how Closehold last stopped (for example, a normal exit, an update or a crash); the Closehold and extension versions; a coarse description of the computer (a range of processor cores and memory, the processor type, a graphics-memory range, and the Windows version); and daily counts of starts and stops, retries, give-ups, switches to another way of protecting a file, and how often and for how long (in ranges) protection was paused. Each problem will be sent with the hour it happened, and normal use will be sent only as daily counts. How many times each AI website or app was used, and the time spent in each step, will be sent with your license key but kept only as totals across all organizations, never against you. Service health never contains what you typed, file content, file names or paths, page titles or addresses, sensitive values or hashes of them, or what kinds of sensitive information were found. We will use it to keep the product working; we will not use it to monitor how you use AI, and we will never sell it or use it for advertising.
An administrator of your organization’s Closehold account can switch Service health off for the whole organization, before it starts or at any time after. On a computer your organization manages, this setting follows the organization. To have the Service health records linked to your license key deleted, email privacy@closehold.ai or ask your administrator.
Account and billing information
If you or your organization sign up for Closehold, we hold the account details needed to run it: your work email address, your organization, your license and the devices enrolled under it, and billing details, which are processed by our payment provider, Stripe. Our servers run on Google Cloud, and we send account email through Resend.
Updates and downloads
The desktop application checks Closehold’s release server for a newer version of itself. That request carries the version it is running and, as every web request does, your network (IP) address. If you choose to download a local AI model for Closehold Clear, it is fetched from Closehold’s storage on Google Cloud in the same way. Neither sends anything you typed or opened.
This website and its forms
closehold.ai has no analytics, no advertising and no tracking cookies. If you close the offer banner, your browser remembers that in its own storage; nothing is sent. Our hosting provider, Google Firebase Hosting, receives each visitor’s network (IP) address in order to serve the page.
The contact form. When you request a demo, early access, our DPA (in preparation), a security review or a sales conversation, ask us to tell you when the macOS app ships, or send us a message, the form sends Closehold’s servers your name, your work email, your company (if you give it), your message and which form you used. We store it on Google Cloud and email a copy to our team, and we use it only to reply to you — for the macOS request, to tell you once, when it ships. We keep it until we no longer need it to reply to you and follow up, or until you ask us to delete it at privacy@closehold.ai, whichever comes first.
Checkout. When you buy a plan on the pricing page, your work email, your organization’s name, the plan, the billing term and the number of seats are sent to Closehold’s servers, which open a checkout page with our payment provider, Stripe. You enter your card details on Stripe’s page; they never reach Closehold. Your account is created only when the payment completes, and from then on it is held as described under “Account and billing information” above. If you leave a checkout unfinished, Closehold keeps nothing from it; Stripe keeps the unfinished checkout under its own privacy policy.
Extension permissions
The extension requests only the permissions its function requires:
- storage — your on/off preferences, the local pairing state, and short payload-free logs of recent notices and protection outcomes, kept in your browser. No content, no prompts, no browsing data.
- nativeMessaging — to pair with the Closehold desktop application on your own machine, which is where content is processed. This is the core of the privacy model.
- declarativeNetRequest — to ensure a file upload to an AI service carries the protected copy rather than the original. Scoped to the supported AI sites.
- alarms — to periodically re-check whether the desktop application is still running. No network activity.
- scripting — when the extension updates, to put its protection back into the AI-site tabs you already have open, so protection resumes without reloading the page. It works only on the supported AI sites listed below.
- Host access (
127.0.0.1, the supported AI sites, and the Gmail / Outlook web mail sites) —127.0.0.1is your own Closehold application, the only place content is sent for analysis. The AI-site access lets the extension read the prompt you are about to send, protect it, and restore values in the reply, on those sites only. The mail-site access exists solely for the opt-in email advisory above and does nothing until you turn that option on.
Data retention
We retain no content from your AI conversations, because none of it reaches us. Data created by Closehold — your preferences, the local reversible mappings, local audit records, and downloaded local AI models — lives on your device until you clear it or uninstall. The diagnostic record on your computer is deleted after 45 days. Uninstalling the desktop application removes the local AI models and reversible mappings it created.
A support report you save and email to us is kept for 90 days from the day we receive it, and then deleted. We use it only to answer you and to find and fix the problem it describes.
Service health is not shared in this release, so no Service health record about you is on our servers. Once a later release sends it, Service health records about a problem (a failure, a refusal, a cancellation, a crash, or a protection that took 3 seconds or longer) will be kept for 45 days, and daily counts linked to your license key for 30 days. After that, only totals across an organization or across all of Closehold will remain, with nothing linking them to you; those will be kept for up to 13 months. Protection analytics are kept for as long as your organization’s account is active and are deleted when the organization asks or closes its account. Account and billing records are kept for as long as the account is active and afterwards for as long as the law requires.
A Secure Send share is kept until it expires (1, 7 or 30 days, as you choose), is revoked, or, for a one-time link, is opened; a stored file is deleted with it, and in any case within 31 days of upload. Our hosting provider’s request logs, which record network addresses, are kept for 30 days.
Security
Processing happens locally, which removes the largest privacy risk — sending sensitive content to a third party — entirely. The browser extension authenticates to the local application with a per-install pairing token, so an arbitrary web page cannot call it. Persisted mappings are encrypted at rest. Closehold is designed to fail toward protection: if it cannot scan content, it does not silently let that content through.
Your choices
- You can turn protection on or off at any time from the extension, unless your organization manages your computer, in which case protection stays on.
- You can clear Closehold’s local data, or uninstall the application, to remove the data it created on your device.
- An administrator of your organization’s Closehold account can switch Service health off for the whole organization.
- Our Data Processing Agreement (DPA) for organizations evaluating or deploying Closehold is in preparation; ask us, and we will send it as soon as it is ready.
- You can ask us what personal data we hold about you, and ask us to correct or delete it, by contacting us using the details below. Your prompts and the files you give an AI tool never reach us, and what you send with Secure Send reaches us only in a form we cannot read. But if you have an account, or your computer is part of an organization, we do hold some: your account details and the protection analytics described above, and, once a later release sends it, the Service health records described above. We will delete the Service health records linked to your license key on request.
Children’s privacy
Closehold is a workplace and productivity tool. It is not directed to children under 16, and we do not knowingly collect personal information from children.
International users
Because core processing is local to your device or your organization’s infrastructure, your prompt content does not cross a border to reach Closehold — it does not reach Closehold at all. Account details, protection analytics and, once a later release sends it, Service health records are payload-free but are personal data; they, and Secure Send shares (which we cannot read), are stored on Google Cloud and may be processed outside your country.
Changes to this policy
We may update this policy as Closehold evolves. When we do, we will revise the “Last updated” date above and post the new version at this address. Material changes will be made clear.
Contact
Questions about this policy, a data request, or a Data Processing Agreement (DPA)? Email privacy@closehold.ai, or use our contact form.