ISSUE 46 · 6 MIN READ ·

Somebody owns noreply.net, and your systems have been emailing him

Placeholder 'no reply' addresses on domains you don't own deliver real mail to strangers. How to find noreply@ addresses in your configs and fix them in ten minutes.

Somewhere in the set-up of a system you rely on, there was a box that had to be filled in. The alerting tool wanted an address to notify. The contact form wanted a “from” address. The new phone system wanted somewhere to send its reports. Nobody had decided what that should be yet, so someone typed the obvious thing — [email protected], say — on the reasonable assumption that it went nowhere. The form saved, the system went live, and the box was never looked at again.

The trouble is that “nowhere” is a real place. noreply.net is a registered domain, and so is noreply.us, and both belong to a security researcher, Cory Solovewicz, who set them up to accept whatever arrived. According to the abstract for his DEF CON 34 talk, he registered the first on a whim in 2020 — and then a city government sent him its internal documents, another organisation started sending him phone-system alerts, and 363,000 emails turned up in sixteen months. Across more than 20 such domains, he has now collected around 400,000 misdirected messages. He never attacked anyone. The mail simply came to him.

This is the quiet failure: a placeholder address that looks like a dead end but is actually somebody else’s inbox.

Why it stays invisible

Email has no concept of a pretend address. When a mail server has a message to deliver, it looks up the recipient’s domain in the Domain Name System (DNS), finds the mail server listed there, and hands the message over. If a domain has no mail record at all, RFC 7505, the internet standard for domains that accept no mail, notes that senders fall back to the domain’s ordinary web address and try that instead. At no point does anything ask whether you meant it. If a stranger owns the domain and has pointed its mail at a mailbox, your message arrives, quietly and successfully.

And nothing on your side looks wrong. There’s no bounce, no error, no angry customer. Alerts that go to a no-reply address were never going to be read, so nobody misses them. Replies to a no-reply sender go unanswered, which is what everyone expected. The only person who knows anything is happening is the one reading your mail.

It also hides because it’s spread across many hands. A developer hard-codes a default into an application; a supplier’s installer ships with one; a marketing tool gets set up by whoever was free that afternoon. Mr Solovewicz’s own summary of the root cause is the assumption “that a domain nobody owns is safe to hardcode” — and plenty of perfectly competent people have made it.

Find it yourself

This is a short hunt, and the aim is simple: a list of every no-reply address your systems use, with the domain next to each one.

  1. Search your code and configuration. If you have developers or access to your repositories, run this from the top of each one: grep -rniEo "(no-?reply|do-?not-?reply)@[a-z0-9.-]+" . | sort | uniq -c. Include the places defaults hide: environment (.env) files, deployment and container configs, and continuous integration settings. You’ll get a count of each address and where it lives.
  2. Check every tool that sends mail on your behalf. Open the settings of your contact and form builders, customer relationship management (CRM) system, helpdesk, newsletter platform, monitoring and alerting tools, phone system, and any scan-to-email printers. Look at the “from”, “reply-to” and “notify” fields. This is also your no-command-line route: it’s slower, but it’s just reading settings screens.
  3. Look at what’s already been sent. Your mail provider’s message trace (in the Exchange admin centre for Microsoft 365, or Email Log Search in the Google Workspace admin console) will let you search for messages sent to addresses containing noreply or donotreply. Anything going to such an address on a domain that isn’t yours is mail leaving the building.
  4. Sort the domains. Anything on your own domain is fine for now. Anything else — noreply.net, noreply.us, a misspelling of your own name, a generic-sounding domain you don’t recognise — needs checking.
  5. Ask who receives mail for each outsider. Run dig +short MX noreply.net (or use any online “MX lookup” page). If it returns real mail servers, somebody is receiving. If it returns 0 ., that’s a “null MX” — the domain’s owner has declared it accepts no mail, and delivery fails straight away. You can see one by looking up example.com, which publishes exactly that. If it returns nothing, don’t assume safety: as above, senders fall back to the domain’s web address.

The usual result is one or two surprises — often an alerting default nobody chose, or a form tool configured years ago by someone who has since left.

The fix

The rule is short: every address your systems send from or to should be on a domain you control.

Move no-reply addresses on to your own domain. Replace [email protected] with [email protected], or with a sending subdomain such as [email protected] if you’d rather keep automated mail separate. As a welcome side effect, mail from your own domain can actually pass Sender Policy Framework (SPF), DomainKeys Identified Mail (DKIM) and Domain-based Message Authentication, Reporting and Conformance (DMARC) checks, since you can publish the records they need — something you can never do for a domain belonging to someone else.

Give it somewhere to land. “No reply” rarely means “no mail”: bounces, out-of-office messages and the occasional customer who replies anyway will all turn up. Point the address at a real mailbox, a shared inbox that someone skims now and again, or a rule that auto-replies with where to get help. Don’t publish a null MX on a domain you send from. RFC 7505 says mail systems should not do that for domains used in sender addresses, and warns that your outgoing mail risks being rejected.

Use null MX for domains that should never receive mail. If you own spare domains, or a subdomain used only for testing, publish a single MX record with priority 0 and a target of ., as RFC 7505 describes. Anything misdirected there then fails immediately instead of sitting in someone’s catch-all.

Use names that are reserved for exactly this. For test data, documentation and genuine placeholders, RFC 2606 reserves .test, .example and .invalid as top-level domains, along with example.com, example.net and example.org. Nobody can register them, which makes them the only truly safe “nowhere”. Something like [email protected] is obviously fake at a glance and can’t reach a stranger.

Stop it creeping back. Add the search from step 1 to your code review or build checks, so a new @noreply.net fails before it ships. When you set up a new tool, make the sending and notification addresses a decision rather than a default, and ask suppliers what their installer puts in those fields. For anything genuinely sensitive already sent to a stranger’s domain, treat it as disclosed: rotate any credentials it contained and tell whoever owns the data.

None of this is difficult. Most of it is finding a handful of boxes someone filled in years ago and changing a few words — which is a modest price for making sure the only person reading your system’s mail is you.

Check your inbox — confirm and you're in. Latest issue: Somebody owns noreply.net, and your systems have been emailing him.

One real, fixable exposure every week. Free.