A browser extension, a Windows tray app, an Android keyboard and an Outlook add-in, all checking on your device. Teams link them to one console.
Browser extension
For Chrome, Edge and Firefox. Checks what you paste, type and attach in AI chats, search boxes and any site you add, then lets you mask it before it's sent.
Stops sensitive data before it's sent to AI chats and search.
Emails, phone numbers, IDs, addresses and more.
Cloud, Git, payment and AI provider credentials.
Spots passwords and connection strings in text.
Checks attached files, not just what you type.
Risk level plus Mask & send, Mask & edit, Edit or Send anyway.
Masks what was found right in the box and lets you keep typing. Nothing is sent until you press Send again.
"My password is…", a PIN or SSN in a sentence, or numbers and keys split with spaces, dashes, dots or hidden characters.
Catches secrets typed straight into the address bar.
Your project code names, client names and internal servers.
People's names and home addresses, matched on your device.
Reads the text in pasted or uploaded images, on your device.
Optional on-device model that also catches names in context, addresses, organisations and health details. Off until you turn it on; nothing is sent.
Blocks the tricks that steal clicks, clipboards and credentials.
Detects invisible overlays hijacking your clicks.
Flags pages that swap what you copy.
Warns when a "verify you're human" page asks you to run commands.
Lookalike domains and AI brand impersonation.
Checks a link before you click it.
Exposes invisible instructions aimed at AI agents.
RDAP age and DNS-over-HTTPS checks are off by default and ask first.
k-anonymity hash prefix only. Off by default, asks first.
Watches what a page actually does, not just its URL.
Optional download, matched locally. Off by default.
Keeps keys, seeds and wallets out of the wrong hands.
Recognises BIP-39 recovery phrases in all 10 official languages (with the checksum) and raw keys.
Unlimited approvals, Permit, setApprovalForAll and blind signing.
Checks risky sites before a wallet is connected.
Fake exchanges, giveaways, airdrops and recovery scams.
Stops adversary-in-the-middle phishing that walks straight past MFA.
Warns and blocks if a password from your real sign-in domains is typed on any other site. Stored only as a local salted hash.
Flags one-time codes entered on look-alike pages that aren't your identity provider.
Catches "enter this code at microsoft.com/devicelogin" lures.
Microsoft, Google and Okta sign-in pages on the wrong domain.
See what was caught and show your working. Generated on your device, in the extension and the Windows app.
Site by category heat-map, 14-day timeline and outcome funnel.
SHA-256 hash chain with one-click integrity verification.
What VeilGuard did and when, kept locally.
CSV, JSON, evidence pack and printable HTML/PDF report.
Keep AI useful without handing over the real data.
The AI sees [EMAIL_1]. You see the real value over it in the answer, on your screen only.
Strips author names, comments and tracked changes from Word, Excel and PowerPoint before upload.
What was caught, where, and tips, built from your local history.
Every warning lets you choose; admins can require masking. Mask & edit is always offered.
One click and a value is never flagged again.
Text, Office and PDF files, saved as a redacted copy.
Right-click: 15 minutes to forever, per site or app.
15 min, 1 h, 3 h, until tomorrow 6:00 or until restart. Shared with the Windows app.
Strict for AI chats, relaxed or off for your own tools.
Built for individuals, ready for teams.
Admins can require masking, so there's no "send anyway".
Manage the Windows app with Group Policy.
Company settings for every device, Microsoft & Google sign-in, admins and view-only staff.
One click for HIPAA, PCI DSS, GDPR, legal, financial services or strict setups.
Counts to Splunk, Microsoft Sentinel or a signed webhook, plus API tokens.
Printable summary with PDF and CSV download.
Checks emails and attachments when you press Send, with one-click masking.
Instant, in-context warnings.
See what was caught, kept on your device.
Put the status where it suits you.
English, Español, Français, Deutsch, 日本語.
Runs quietly in the background.
Every pattern is published.
Stops git commits that add secrets (Windows app).
Windows tray app
Sits in the system tray and checks what you paste or send in Claude desktop, VS Code, Microsoft 365 Copilot, Word, Outlook, Teams, Cursor and ChatGPT desktop, by intercepting paste and Enter.
Installs for your user only. No admin rights, no .NET needed. IT can roll it out silently.
Reads the text in pasted images with the OCR engine built into Windows, on your PC.
Checks every git commit on your PC and stops one that adds keys or passwords, showing the file and line. Your repositories' own hooks still run.
Word, Excel, PowerPoint, PDF and text files, saved as a redacted copy.
Holds Word, Excel, PowerPoint and PDF files with a Microsoft sensitivity label you protect, such as Confidential, before they are pasted or uploaded.
Optional: checks files copied to USB drives and documents you print. Monitor, warn or block. Block removes the copied file from the drive and cancels the print job, since a Windows app can't stop the copy itself.
Reads admin policy from the registry and ships ADMX templates, including forced redaction.
Shares settings, history and the company link. The extension asks the app to read images, so they never go to a server.
When a send is paused, mask what was found and keep typing in the same app. Nothing is sent until you press Enter again.
Pair VeilGuard Keyboard with a QR code and copy text between phone and PC. Local network only and end-to-end encrypted, with nothing going through our servers. Anything VeilGuard flags is held until you choose.
Age-appropriate checks for prompts in desktop AI apps such as ChatGPT, Claude and Copilot, behind a parent PIN. More below.
VeilGuard Keyboard for Android
Phones are where most slips happen. VeilGuard Keyboard is a full Android keyboard that checks what you're about to send in any app, on the phone itself, and lets you mask it before it leaves.
Cards, passwords, API keys, national ID numbers from 29 countries, health and bank details, and look-alike phishing links. Caught even when written in a sentence ("my pin is…") or split with spaces. Mask & send, mask & edit, use placeholders, edit or send anyway.
Copied images and screenshots appear in the clipboard history and as a paste suggestion. VeilGuard reads the text in them on the phone and warns before a card or password in a screenshot is pasted. Location and camera details are removed.
Copy on the phone, paste on Windows, and back. Paired by QR code with the VeilGuard Windows app, over your local network only and end-to-end encrypted. Off until you turn it on.
Suggestions that learn your style, phrase completion, autocorrect and grammar fixes, glide typing, voice typing with spoken punctuation, handwriting, Morse code, emoji, stickers and GIFs you can swipe through and search, clipboard history, password-manager suggestions and a space-bar trackpad.
61 languages including Chinese (Pinyin), Japanese (kana-kanji) and Korean, two at once if you like, Hindi, Russian, Arabic, Greek and more typed in Latin letters, right-to-left support, Dvorak, Colemak and national layouts, a key mapper to change any key, split, one-handed and floating modes, themes, your own colours and photo backgrounds. Works with TalkBack and physical keyboards.
Every check runs on your phone. Incognito typing in one tap. Your personal dictionary and clipboard history are encrypted on the device, and backups are passphrase-encrypted files you keep. Translate, GIFs, handwriting, web search and AI rewrite are optional, each with its own switch, and text is checked before anything leaves the keyboard.
Link to your company console, or set rules through Android Enterprise managed configuration, including one-switch strict mode for regulated teams.
Grab text from the camera, checked before it's inserted. Optional on-device AI also catches names and addresses.
Age-appropriate checks on send in AI apps, browsers and search fields, behind a parent PIN. More below.
Outlook add-in
The subject, message, recipients and attachments are checked for personal data, card and bank numbers, API keys, passwords and your company's own terms. The alert names what kind of data and how much, never the data itself.
Illustration of the send alert. It names the kind of data and the count, not the values.
Outlook on the web, new and classic Outlook for Windows, and Outlook for Mac.
Your IT admin deploys it from the Microsoft 365 admin center (Integrated apps), with the send check on for everyone.
Checks run inside Outlook; nothing in your mail is sent to us. Teams: the Windows app checks the Teams desktop app at send.
Team console
Link every browser, PC and work phone to your company ID. Your admin sets the rules at veilguard.site/console. Only counts and event details (when, which app, what kind of data) are shared, never the text, files or previews.
Policy templates available in the console, one click each.
Company settings pushed and locked: protected sites and apps, packs and custom rules. Masking can be required.
Lets someone change a setting when they need to.
Microsoft & Google sign-in, multiple admins and view-only staff.
Counts of what was caught on each machine and the list of protected machines.
Every “send anyway”, block and forced redaction in one list: acknowledge, resolve, assign, add notes, export CSV. Optional short reason from the user.
Which AI tools your team uses, per day and device, how many prompts had findings, and which tools are not approved.
Counts to Splunk, Microsoft Sentinel or a signed webhook, plus API tokens.
Printable summary with PDF and CSV download.
Discord, WhatsApp Web, Slack, Teams, Telegram and Messenger protected by default.
Android Enterprise managed configuration for the keyboard.
Group Policy, Intune and browser policy roll-out on Enterprise.
Optional on-device AI detection
An optional on-device model that also catches names in context, addresses, organisations and health details, on top of the published rules. It's off until you turn it on, it runs locally in the extension, the Windows app and the keyboard, and nothing is sent.
Family protection
Choose an age (under 13, 13 to 15 or 16 to 17) and VeilGuard stops AI prompts and web searches that aren't right for it before they are sent. It works in ChatGPT, Claude, Gemini, Copilot, Grok and other AI chats, in search boxes and in searches typed into the address bar, in the desktop AI apps on Windows and on VeilGuard Keyboard.
Sexual and explicit content, self-harm and suicide methods, eating-disorder promotion, weapons and explosives, illegal drugs, buying alcohol, tobacco or vapes, gambling, hookup sites and stranger chat, gore, extremism, dangerous challenges and ways around parental controls. Younger ages also skip dating apps, mature topics and horror.
Puberty, sex education, consent, contraception, how to report abuse, bullying, mental health, the risks of drugs, history and the news all go through. Tricks like p0rn, spaced-out letters and hidden characters are caught, in English, Spanish, French and German.
A short, friendly message instead of the answer, with no way around it for the child. If the question is about hurting themselves, VeilGuard shows a caring message with the 988 Suicide & Crisis Lifeline (US), Childline 0800 1111 (UK) and findahelpline.com, and asking for help is never blocked.
Only a parent PIN can change the age or turn it off. A parent can unlock it for 15 minutes or an hour. Sexual content involving minors stays blocked for every age, even when unlocked.
Every check runs on the device. Nothing is sent anywhere, and the history keeps only the topic and the time, never what was typed.
Set the age for every device from the team console, browser policy, Group Policy or Android managed configuration. Reports count blocks by topic only.
A helperFamily protection makes it much harder to ask for things that aren't right for a child's age, but it is not a guarantee: no filter catches everything. Keep using your device's and browser's own parental controls, and keep talking with your child.
Built for regulated teams and everyday users alike, in the browser extension and the Windows app. Still 100% on your device.
Right-click VeilGuard and choose Redact a document. Pick a text, CSV, JSON, Word, Excel, PowerPoint or PDF file: VeilGuard replaces every sensitive item and saves a copy as name_redacted_date-time. PDFs are rebuilt with the text blacked out, so it can't be recovered.
LocalThe file never leaves your PC.
Dismiss it, or snooze it on that site or app for 15 minutes, an hour, a day, a week or for good. Snoozes are listed in Settings, where you can undo them.
Or pause the send, paste and upload warnings on a whole site (or app) for 15 minutes, 1 hour, 3 hours, until 6:00 tomorrow or until restart, from the warning, the toolbar popup, the tray menu or Settings. Security warnings pause only if you tick the box; companies can turn pausing off or cap its length, and forced redaction is never paused.
UpdatesThe extension shows its version and checks for updates.
Placeholders become [PERSON_1] and [ADDRESS_1]. Names are matched against public first-name and surname lists, on your device.
Banking & financial services, government, military & defence, education, insurance & HR, and telecoms: from CVV codes and classification markings to student records, ITAR text and SIM numbers.
171detectors in total, including national IDs from 29 countries.
Rules 1.4.2 catch passwords, PINs, SSNs, card and bank numbers, IDs, dates of birth and API keys typed in plain sentences ("my openai key is…", "mot de passe est…"), spaced out, split with dashes or dots, or hidden with invisible characters, in English, Spanish, French, German and Japanese. Phrases like "password is incorrect" stay quiet.
One click tells VeilGuard a value is fine. Set each site or app to strict, normal, relaxed or off.
The Windows app can stop a git commit that adds keys or passwords, showing the file and line.
A first-run tour shows what the border means and lets you try VeilGuard with test data. The test page checks secrets, fake CAPTCHAs and searches in one minute.
Every pattern VeilGuard checks for is published here, with the exact rules file the apps use. See what is matched, and that nothing is sent.
Free for individuals, forever. Add the rest whenever you like.