Help

TempMail or a permanent mailbox

extort.fans offers two routes, and they do not exclude each other.

A TempMail address is already waiting when you open the home page. No sign-up, no password, no confirmation. It is meant for sign-ups that demand an address but do not justify a lasting relationship: a download behind a form, a forum, a trial account. Once it expires it disappears completely.

A permanent mailbox is a full mail account: your own address, a password, 2 GB of storage, folders, archive, trash. It can also send, which a TempMail address never can. Reading and writing happen in the webmail — there is no IMAP access, for a reason spelled out at length under Permanent mailbox.

If you are unsure: take the disposable address. It costs you nothing and commits you to nothing. If you later decide you want to keep it, one click turns it into a permanent mailbox without losing the messages already received.

Creating a mailbox

Enter your preferred address on the right of the home page. As you type, the page checks availability; a tick in the field means free, a cross means taken or reserved. Three to thirty-two characters are allowed, from lowercase letters, digits, dot, hyphen and underscore.

Then a password of at least twelve characters, twice. This password is more than a login: the key that decrypts your messages is derived from it. The server never sees it and cannot reset it.

Right after creation, extort.fans shows you your recovery kit once: twenty-four words. Write them down or print them. They are the only way back if you forget your password. To confirm, the screen asks for three random words. After that you land in the inbox.

No confirmation mail is needed. You can add a recovery address later in the settings.

Receiving your first mail

Copy your address — one click on the address is enough — and enter it somewhere that someone will write to you. New messages appear without a reload, both on the home page below the two cards and in the mail client.

If nothing arrives, wait a minute: some senders deliberately delay the first delivery. If it stays empty, the Troubleshooting section helps.

The mail client

The actual workspace lives at /mail: folder list on the left, message list in the middle, reading view on the right. Column widths can be dragged and are remembered.

If you prefer the keyboard: c compose, r reply, a reply all, f forward, e archive, j and k navigate, / search, ⌘K or Ctrl+K opens the command palette. Pressing ? shows the full list.

Creating an address

The address is created when you open the home page, without a click. The local part is ten random characters from an alphabet that leaves out confusable glyphs: no i next to l, no 0 next to O. That way an address can be read aloud or typed out without anyone having to guess.

The random numbers come from the browser’s cryptographic random generator. There is no counter logic and no timestamp in the name that anything could be derived from.

Lifetime and extension

The default is sixty minutes. The countdown next to the address runs in plain sight. +1 h adds another hour, as often as you like, up to the ceiling of seven days from creation. Once the ceiling is reached the interface says so and the button stops doing anything.

You can also pick the lifetime up front: ten minutes, one hour, six hours, twenty-four hours or seven days. Ten minutes is enough for a one-off confirmation mail. Seven days suits a trial account over a weekend.

A custom prefix

Under Customise the address you can set your own local part instead of the random sequence, along with domain and lifetime. Availability is checked while you type. Reserved names such as postmaster, abuse, admin or noreply are blocked, as are addresses that expired within the last thirty days.

A custom prefix is convenient, but guessable. Anyone choosing [email protected] has to expect that somebody else had the same idea and the address is taken — or that somebody requests it deliberately after it expires. For anything that matters, the random address is the better choice.

Why a TempMail address cannot send

Disposable addresses without sign-up are, if they are allowed to send, a tool for spam, fraud and extortion. So sending is not switched off but ruled out structurally, in two independent places.

First in the API: every send attempt with a TempMail token is rejected before anything happens. Second in the mail server itself: Postfix accepts no submission for relaying from these addresses, because they have no SASL account and relaying requires authentication.

That covers replies and forwards too. There is no switch that changes it, and no exception for signed-in users who happen to have a TempMail address running on the side. Anyone who wants to send needs a permanent mailbox.

What happens at expiry

When the countdown reaches zero, a cleanup run starts:

1. The ciphertext of every message is deleted from disk, attachments included.
2. The address’s public key is removed.
3. The alias is deactivated; further incoming mail is rejected.
4. The metadata is reduced to plain counters.
5. The address goes onto a block list for thirty days.

The block period has a reason: without it, somebody could request a just-expired address straight away and receive the mail still meant for you. After thirty days the address is free again.

None of this can be undone. There is no trash window and no copy in a backup, because a backup that could restore deleted disposable mail would be precisely the opposite of what a disposable address is for.

The address on a second device

The private key of your TempMail address lives in the browser, in IndexedDB, and does not leave the device. Another device therefore cannot read the messages, even if it knows the address.

If you want to use the address on your phone, use the QR code: it carries the address so you can enter it where it is needed. Reading still happens on the device that holds the key.

If you want to read messages on several devices, a permanent mailbox is the right fit. There the key hangs on the password, not on the browser.

Limits

• At most three new addresses per hour and internet connection.
• At most fifty messages per address.
• Attachments up to ten megabytes, scanned for malware before encryption.
• No folders, no aliases, no sending.

Sign-up

Enter an address and a password on the home page, create it, done. No confirmation link, no phone number, no payment details. Reserved addresses are blocked; available ones are marked as free while you type.

The password must be at least twelve characters. Length beats special characters: a sequence of four random words is stronger and easier to remember than P4ssw0rd!. The strength meter under the field weighs length first.

Recovery kit

Immediately after creation the recovery kit appears: twenty-four words following the BIP39 scheme. Your master key can be reconstructed from them if the password is gone.

You can download and print it. To confirm that it really is saved, the screen asks for three random words. The kit can be shown again later in the settings, as long as you are signed in.

Without the password and without the recovery kit the mailbox is lost for good. That is not customer-service hardness but the direct consequence of the server never holding your key. There is no administrator access that could get around it.

Do not keep the kit in the same password manager as the password. Paper in a folder is a seriously good solution here.

Why there is no IMAP access

Thunderbird, Apple Mail and Outlook do not work with extort.fans. That is not a missing feature but the direct consequence of the encryption — and because it often comes as a surprise, here is the full reasoning.

Your messages sit on disk encrypted, against a key derived from your password in your browser. The server does not hold it. An IMAP client speaks the IMAP protocol and knows nothing of this scheme; it would be handed encrypted data by the server and would display unreadable noise.

There would be three ways out, and each costs more than it gives:

The server could hold the key as well. Then IMAP would work — but zero-access protection would be lifted for your entire mailbox, not just for the messages read over IMAP. That would remove the reason to use this service at all.

There could be a local bridge program that decrypts on your machine and offers IMAP itself. That is how Proton Mail solves it. It is a separate program with its own installer for every operating system — a project in its own right.

Or there is no IMAP. That is the path we chose.

The webmail at /mail is therefore the only way in. It works in every current browser, on the phone too, and needs no installation. Anyone who absolutely needs a local client is better served by a provider without zero-access encryption — saying that is more honest than an access route that delivers gibberish.

Aliases

Up to ten additional addresses delivering into the same mailbox. Useful for keeping senders apart: shop@, club@, newsletter@. Each alias can be switched off individually without the already delivered messages disappearing.

When composing, you can choose which address appears as the sender.

Filters and out-of-office

Filters are clicked together in the interface and generated behind the scenes as Sieve rules that run server-side — so they work even when no browser is open.

Conditions: sender, recipient, subject, content, size, attachment present. Actions: move, label, flag, forward, delete, mark as read. The order can be dragged; the first matching rule wins unless you set it to continue.

An out-of-office notice with a date range sits next to it. It replies to any given sender at most once a week.

Storage

2 GB per mailbox. The bar in the sidebar shows the level. At 90 per cent a notice appears; at 100 per cent incoming messages are rejected rather than silently dropped — the sender gets an error and knows about it.

Attachments are the biggest item. Searching for larger:5M finds the fattest ones.

Import and export

Export as an mbox file. The messages are decrypted in your browser for this and downloaded as a plaintext archive — afterwards it sits unencrypted on your disk, which is the whole point of an export. Put it somewhere you control.

Deleting the account

At the very bottom of the settings. Deletion requires the password and a typed-out confirmation. Afterwards the mailbox, ciphertext, attachments, search index, aliases and sessions are removed.

The address goes onto the block list for thirty days so that it is not handed straight back out and starts receiving somebody else’s mail. Nothing about this can be undone.

What the operator can see — and what not

Honesty matters more than marketing here, so both sides in plain words.

Not visible: the content of your messages, the subject lines, the preview texts, the file names of your attachments, their content, your notes on contacts and your search index. All of that sits on disk as ciphertext only. The server does not hold the matching key and cannot derive it.

Visible: that a mailbox exists, when it was created, how many messages it holds and how large they are, who wrote to whom and when, which network you sign in from and which folders you created. This metadata is needed for delivery, spam defence and abuse handling. Encrypting it would mean no longer being able to deliver mail.

Anyone who needs metadata protection needs a different tool than email. That is not a quirk of extort.fans but of SMTP.

How it works

Argon2id derives a key from your password in the browser: the key encryption key. That decrypts your master key, from which your PGP key pair and the mailbox content key follow.

On the server there is only: the salt, the encrypted master key, your public key and your encrypted private key. Never the password, never the key encryption key, never a key in the clear.

Sign-in is separate from this. It runs over SRP-6a, a scheme in which the server is shown proof that you know the password without learning it. The verifier on the server is of no use for decrypting your mail.

Incoming messages are encrypted against your public key before they are written to disk. The plaintext exists only briefly in the delivery process’s memory and is overwritten there afterwards.

Decryption happens in your browser, in a web worker, so the interface does not stall.

On top of that, the volumes for Maildir and object storage sit under LUKS2, and backups are encrypted with age. That protects against the stolen server, not the running one — which is why it is an addition, not a replacement.

What it protects against, and what it does not

Protected: somebody pulls the disk out. A backup falls into the wrong hands. A database dump ends up online. An authority seizes the server. In all these cases the attacker gets ciphertext and metadata, no content.

Not protected: your own device. Anyone who controls your machine or manipulates your browser reads along the moment you decrypt. Equally unprotected is the case where somebody with control over the server ships manipulated JavaScript — that is the fundamental weakness of all browser cryptography and applies to every provider of this kind. And metadata stays metadata.

Anyone who has to be secure against these cases encrypts with a local PGP program before the message leaves the machine.

Why search runs in the browser

A server-side full-text search would need to be able to read the content. That would give up zero-access protection — not theoretically, but in everyday use.

Instead the search index is built in your browser, maintained in a web worker and stored encrypted on the server so that your second device does not have to rebuild it. The server keeps a blob it cannot read.

The price: the first search after signing in on a new device takes a moment, until the index is loaded and decrypted.

A lost password

With the recovery kit: enter the twenty-four words, set a new password, carry on. The master key is reconstructed from the kit and re-encrypted against the new key encryption key. Nothing is lost.

Without the recovery kit: nothing can be done. No support ticket, no photo ID, no procedure. The key exists nowhere outside your head and your kit. A way back that the operator could open would be a way somebody else can take too.

Checking a fingerprint

Clicking the sender opens a popover with the name, address and, where available, the fingerprint of their key. A fingerprint is the short form of a public key.

Checking it only becomes meaningful once you compare the fingerprint over a second channel: read aloud on the phone, matched in person. If it agrees, you know the message really came from the holder of that key. If it changes unexpectedly, caution is in order.

Sending to recipients without PGP

Most recipients have no PGP key. extort.fans then offers two routes.

A password-protected link, the default. The message is stored encrypted, the recipient gets a link, and you give them the password over another channel. The link expires and has a limited number of views.

Unencrypted sending. Perfectly ordinary mail, as it arrives everywhere, with transport encryption between the servers and no protection on the target system. Sometimes that is exactly right; the interface asks once and does not remember the decision.

Two-factor sign-in

In the settings under Account. extort.fans uses time-based one-time codes following TOTP, as generated by Aegis, 1Password, Bitwarden or Google Authenticator. There is no SMS, because a phone number can be taken over.

During setup a QR code appears, and below it ten backup codes. Each code works once. Save them together with the recovery kit, but not on the same device that runs the authenticator app.

Worth noting: two-factor protects the sign-in, not the encryption. Anyone who knows your password and gets at your encrypted data can decrypt it without the second factor too. The second factor stops somebody with a stolen password from signing in to your mailbox and reading, sending or deleting mail.

Active sessions

The settings list every signed-in session with its time, approximate location by IP address and browser identification. Anything you cannot place, you sign out there; the session is invalid immediately.

Have a quick look after a trip or after using somebody else’s machine. If you find an unknown session: sign it out first, then change the password, then reissue the app passwords.

Blocked images

Remote images in HTML mail are the most reliable counting mechanism senders have. An invisible one-by-one pixel reveals that you opened the message, when, how often and from which network.

extort.fans therefore does not load remote images automatically. A slim bar above the message offers to load them if you want. For senders you trust, this can be allowed permanently.

Read receipts are switched off across the board and are never sent.

Spotting phishing

Most attacks need no technical hole, just haste.

The display name is worthless. Anyone can call themselves “Bank Security”. Click the sender and look at the actual address, especially the part after the last dot before the @. [email protected] belongs to example.tld, not to the bank.

Check links before clicking. A link’s text says nothing about its target. Hover over it and read the address at the bottom. Watch for lookalike domains: an rn instead of m, one hyphen too many, an unfamiliar suffix.

Time pressure is the warning sign. “Your account will be locked in 24 hours” is not information, it is a method. No legitimate sender needs your reaction within minutes.

Attachments you did not expect. Especially Office files with macros and archives containing a single executable. When in doubt, ask the sender over a known channel, not via the message’s reply address.

Nobody from extort.fans asks for your password. Not by mail, not on the phone, not in chat. Whoever does, is not us. And since the operator does not hold your key anyway, there would be no reason for it either.

What to do if something has happened

1. Change the password. This re-encrypts the master key against the new key.
2. Sign out all foreign sessions.
3. Revoke all app passwords and reissue them for the devices you actually use.
4. Review the filter rules. A favourite trick is an inconspicuous forwarding rule.
5. Check the aliases for a new one that has appeared.
6. Turn on two-factor if you have not already.

Anything suspicious can be submitted under Report abuse.

A mail does not arrive

In order, which saves time:

Is the address still valid? For TempMail the countdown shows the time left. If it says “expired”, the address and its contents are deleted and it accepts nothing more.

Is the spelling right? Random addresses avoid confusable characters, but a self-chosen prefix with a dot or hyphen is easily mistyped. Copy the address rather than transcribing it.

Has the message limit been reached? A TempMail address takes fifty messages. After that it rejects.

Is the message too large? Attachments over ten megabytes are refused for TempMail, and the sender gets an error.

Did the sender get an error message? That is the most valuable piece of information. A message saying 550 5.1.1 mailbox unknown means the address does not exist here — usually a typo or an expired address. Domain not found means the domain name was misspelt.

Some senders delay deliberately. One to two minutes is normal, especially on first contact.

A mail lands in spam

Incoming messages are scored by Rspamd. Above a threshold they land in the spam folder rather than being discarded — so you can always go and look.

Move a wrongly sorted message into the inbox. That trains the filter for your mailbox. Moving into the spam folder trains it the other way.

If wanted mail from a sender regularly lands in spam, create a filter: condition sender, actions “move to inbox” and “mark as not spam”.

TempMail addresses have no folders. There, messages with a very high spam score are rejected outright and everything else lands in the inbox.

Sign-in fails

“Wrong password” although it is right. Check caps lock and the keyboard layout. Special characters sit elsewhere on an English layout. Copy the password from your password manager instead of typing it.

The two-factor code is rejected. Usually the phone’s clock is off. TOTP codes hang on the time; even a minute of drift is enough. Turn on automatic time synchronisation on the phone and try again. Otherwise use a backup code.

After signing in the inbox stays empty and spins. Then decryption is running, or failing. Reload the page. If it persists, browser storage may be blocked on this device — see below.

Forgotten password. Only with the recovery kit. Without the kit there is no way back, see Encryption.

An attachment will not open

Attachments are decrypted in the browser and then downloaded. If that breaks off, this usually helps:

• Do not start several downloads in parallel; start one and wait.
• With very large attachments, leave enough free memory; decryption briefly needs roughly twice the file size.
• In a private window or with strict blockers the download can be prevented. Try once in a normal window.

PDFs and images can be viewed directly in the reading view, without downloading.

Browsers

Web Crypto, IndexedDB, web workers and server-sent events are required. Every current browser provides these.

Tested: Firefox, Chrome, Edge and Safari, each in the current and previous major version.

Not supported are Internet Explorer and browsers with JavaScript fully switched off. The home page is server-rendered and readable without JavaScript, but the disposable address and the mail client need it.

Private window. Works, but the key for your TempMail address sits in IndexedDB and is deleted when the window closes. The address then still exists, but its messages are no longer readable.

Blockers and strict privacy settings. Blocking storage access for the site means losing the session on every reload. extort.fans loads no third-party scripts and sets no tracking cookies; an exception for this site costs you nothing.

The service does not respond

The current state of all subsystems is at Status. The page refreshes every thirty seconds and shows whether the database, cache and live connection are running.

If everything says “up” and it still jams, the problem sits between you and the service. Try another network, then report it via Abuse and contact.

The terms in short

The full version is at Terms of use. The essentials in plain words:

The service is free and runs without any availability commitment. There is no guarantee that a message arrives, and none that the service is still running tomorrow.

Anything lawful is allowed. Not allowed, in particular: sending unsolicited bulk mail, fraud and phishing, distributing malware, harassment and threats, distributing depictions of child sexual abuse, and any attempt to attack the service or third parties.

Breaches lead to immediate suspension without warning. Criminal offences are reported to the authorities.

Disposable addresses technically cannot send. That makes them useless as a tool for spam and extortion, and that is deliberate.

Reporting abuse

Through the form at Abuse or by mail to [email protected].

Useful for handling a report: the address involved, the time of the message, the full mail header — in the mail client under the menu entry “Original headers” — and a short description.

Reports are reviewed. Confirmed abuse gets the mailbox suspended and the ciphertext deleted. The reporting person only hears back if a contact address was given.

Law enforcement requests

Requests from law enforcement are handled when they come in the form applicable law requires. What can be handed over is limited by the technology:

Can be handed over: the metadata that operations produce anyway — that a mailbox exists, when it was created, sender and recipient addresses, timestamps, sizes, IP addresses from the connection logs within their retention period.

Cannot be handed over: content, subject lines, attachments, contact notes and search indexes. They exist as ciphertext only, and the key does not exist on the server. An order changes nothing about that, because there is no procedure by which the operator could make this data readable.

Not possible: building a back door that retroactively makes already stored messages readable. Anyone with access to an endpoint device or to future shipped code can bypass browser cryptography — that point is stated openly in the threat model and is not glossed over here.

What is stored and for how long

Data	Retention
Messages in permanent mailboxes	until you delete them
Deleted messages in the trash	30 days, then permanently
TempMail messages	until the address expires, then immediately and completely
Blocked TempMail addresses	30 days, the address name only, no content
Mail server connection logs	7 days
Web server access logs	7 days, IP addresses truncated
Active sessions	until sign-out or expiry
Audit log for administrative actions	90 days, without message content
Deleted accounts	removed immediately, address blocked for 30 days

There is no usage tracking, no analytics, no ad networks and no sharing with third parties for advertising. The pages load no external scripts, which is why no consent banner is needed either.

Full version: Privacy.

Responsible party

Details about the operator are in the legal notice.