Email authentication in practice: what 51 state governments get wrong
92% of state government portals publish a DMARC record. Only 29% enforce it. Here is what that gap means, how SPF, DKIM and DMARC fit together, and how to check a domain in about a minute.
We scanned the 51 U.S. state and District of Columbia government primary portals to see how email authentication is actually deployed, rather than how it is recommended. The result is a useful lesson in the difference between publishing a policy and enforcing one.
- 47 of 51 publish a DMARC record (92.2%).
- 15 of 51 enforce it with
p=reject(29.4%). - 21 of 51 sit at
p=none, which requests no action at all. - 4 of 51 publish no DMARC record.
- 51 of 51 publish SPF, but only 27 use a hard fail (
-all).
Those numbers are from a single snapshot dated 5 October 2026, and they are the whole argument of this guide. Authentication records are easy to publish and easy to publish badly.
What the three records do
SPF is a list of hosts allowed to send mail for your domain. It lives in a TXT record at the apex and is checked against the envelope sender. It breaks when mail is forwarded, because the forwarding server is not on your list.
DKIM signs messages with a private key and publishes the matching public key at a selector under _domainkey. It survives forwarding because the signature travels with the message. It fails silently if the selector is wrong, the key is rotated without republishing, or a mailing list rewrites the body.
DMARC ties the two together. It tells a receiving server what to do when a message claiming to be from your domain fails both checks, and it asks for reports either way. Without DMARC, SPF and DKIM failures are advisory.
The p=none problem
A p=none policy is a monitoring mode. It asks receivers to deliver failing mail as normal and send you a report. That is the correct way to start, because it shows you what legitimate mail would break before you tighten anything. It is not a destination.
Twenty-one state portals are still sitting in that mode. Their DMARC record exists, their reports are probably arriving, and a message that spoofs them is still delivered. If a domain has been at p=none for more than a few months, the monitoring has either stopped being read or nobody owns the decision.
Moving to p=quarantine first and p=reject later is the cautious path. Reject is the only policy that stops spoofing outright, and it is what 15 of 51 portals now run.
SPF is universal and still weak
Every portal we checked publishes SPF. That is the easy part, and it is also where expectations go wrong. A soft fail (~all) marks a message as suspicious and lets it through. A hard fail (-all) tells the receiver to reject it. Twenty-seven portals use -all; twenty-one use ~all and one uses ?all, which asserts nothing.
Two domains publish an SPF record with no all mechanism at all. That is a record that exists to satisfy a checklist, not to do anything.
How to check a domain
- Look up the DMARC record at
_dmarc.<domain>. Note the policy, the subdomain policy, and the reporting addresses. - Look up SPF at the apex. Find the
allmechanism and note whether it is-,~or?. - Check the DKIM selectors your provider uses. Common ones are
selector1,selector2,google, anddefault. A missing selector is not proof of no DKIM, only of a selector you did not guess. - Send a test message to a service that returns the authentication results, and read the headers yourself.
Our DMARC and SPF checker runs the first two steps and reports the policy verbatim, which is usually enough to tell whether a domain is enforcing anything.
Worked example: two records, two postures
Two domains, both with impeccable reputations, both with records that look similar at a glance.
$ dig +short TXT _dmarc.google.com
"v=DMARC1; p=reject; rua=mailto:[email protected]"
$ dig +short TXT google.com | grep spf1
"v=spf1 include:_spf.google.com ~all"
Google enforces, but their SPF ends in ~all, a soft fail. That looks like the mistake described above and it is not, which is the part worth understanding. DMARC passes if either SPF or DKIM aligns with the From domain. Google signs everything with DKIM, so the weak SPF qualifier almost never decides the outcome. Aligned DKIM is doing the work; SPF is a backstop.
Now a domain that exists only to be an example:
$ dig +short TXT _dmarc.example.com
"v=DMARC1;p=reject;sp=reject;adkim=s;aspf=s"
$ dig +short TXT example.com | grep spf1
"v=spf1 -all"
That is the maximally locked-down shape. sp=reject extends the policy to subdomains, adkim=s and aspf=s demand strict alignment rather than relaxed, and -all says no host may send as this domain at all. For a domain that never sends mail, that is exactly right. For one that does, it would be a self-inflicted outage.
The lesson from the pair: read the qualifier and the policy together, and check DKIM before concluding anything from SPF alone. A ~all with strong aligned DKIM is a different posture from a ~all with no DKIM at all, and only one of those is a problem.
What to do with the answer
If the policy is p=none, the domain is not protected against spoofing no matter how good the record looks in a screenshot. If SPF ends in ~all, failures are advisory. Neither is a failure to configure anything, exactly. Both are a failure to finish.
Before tightening either, collect a few weeks of DMARC reports. They list the sending hosts and the authentication results, which is the only reliable way to find the application on a forgotten server that sends as your domain. Then move to quarantine, then to reject.
The full dataset, including per-state rows, is on the state government security study page. Method: public DNS records plus one TLS handshake per host, run on 5 October 2026 against the 51 state and DC government primary portals. No port scanning, no vulnerability probing, no crawling. Figures describe that snapshot and nothing later.
Keep reading
- TLS and DNS in practice: what a public audit of 51 state portals showsEvery state government portal answers on 443. Only two thirds negotiate TLS 1.3, and a quarter validate DNSSEC. Here is what that means and how to check your own domains.
- Reading the results: IPs, subnets, ports, and WHOISWhat each lookup actually tells you, and how to interpret the answer. A practical reference for IP addresses, subnet math, open ports, and registration records.