[ DMARC & policy ]

Internationalised email identities: compare DNS forms without rewriting the message

Preserve Unicode and ASCII domain representations, distinguish the mailbox local part from the domain, and apply the relevant conversion before DMARC discovery.

By the EmailAuth research team · 4 min read

[ Key points ]

  • Preserve Unicode and ASCII domain representations, distinguish the mailbox local part from the domain, and apply the relevant conversion before DMARC discovery.
  • Representation conversion is not a rewrite of signed message content.
On this page

An internationalised sender can look different in a mail client, a raw header, a DNS tool and an authentication result. A correct investigation explains those representations before deciding that the identities differ.

Do not compare only the text rendered on screen. Keep the raw header and the extracted domain, identify the applicable internationalisation rules and record the DNS form used for policy discovery and identity comparison. A conversion used for evaluation is a different operation from rewriting a signed message.

Separate three parts of the address

A display name is the human-facing label. The mailbox local part is the portion before @. The domain is the portion after @. They have different rules and different roles in authentication.

Internationalised email standards allow supported systems to transport UTF-8 header content and mailbox names using SMTPUTF8. IDNA addresses internationalised domain labels; it does not turn a Unicode mailbox local part into a DNS name.

DMARC's central identity is the Author Domain extracted from From. Changing the spelling of a person's local part does not create a new DMARC alignment mode, and authenticating the domain does not certify the person or the meaning of the display name.

Keep the Unicode and DNS representations together

IDNA distinguishes Unicode U-labels from their ASCII A-label representation. A valid A-label can be recognised by its xn-- prefix, but a prefix alone does not prove that an arbitrary string is a valid encoding. Use an implementation of the applicable IDNA rules rather than a hand-edited substitution.

RFC 9989 Section 5.3.1 requires conversion of a U-label Author Domain to A-label form for further DMARC processing. That means a visually different header and DNS name may represent the same domain. It does not mean that similarly shaped Unicode characters in two unrelated names become equivalent.

The following example is synthetic and uses the reserved .example namespace. No DNS publication, message delivery or receiver result is claimed.

  • Item: From mailbox with Unicode domain; Illustrative value: info@bücher.example; Purpose: Raw internationalised header identity
  • Item: Corresponding A-label domain; Illustrative value: xn--bcher-kva.example; Purpose: Domain form for the DMARC processing described here
  • Item: Direct DMARC lookup name; Illustrative value: _dmarc.xn--bcher-kva.example; Purpose: TXT query for that Author Domain
  • Item: Equivalent ASCII-domain mailbox form; Illustrative value: info@xn--bcher-kva.example; Purpose: Same domain representation in ASCII form
  • valid passing DKIM identity using that same A-label domain can be compared with the converted Author Domain. If they are identical, the domain comparison can meet strict alignment. A complete DMARC pass still requires a valid applicable policy and a passing aligned authentication path; the conversion itself is not authentication.

Do not apply domain encoding to the local part

Consider a hypothetical mailbox with a Unicode local part at the same internationalised domain. The domain can have an A-label representation, while the local part remains a mailbox identifier governed by internationalised email handling. Applying Punycode to the whole address is the wrong model.

Check SMTPUTF8 support for the actual transport path when the message requires it. A domain written entirely in ASCII A-label form does not, by itself, require SMTPUTF8; other address or header content may still do so. A UTF-8 raw header or non-ASCII mailbox local part introduces a separate transport and parser compatibility question.

Keep these outcomes separate: inability to transport an internationalised message, failure to parse its identity, failure to discover policy and failure of authentication are different stages. A change to a DMARC tag cannot solve all of them.

Preserve signed content during investigation

Retain the original message bytes and relevant signatures. Converting an extracted domain for a comparison does not require replacing the From header in the original message. Rewriting the header and then verifying the modified copy can produce a different signature question.

Record the library or product version performing IDNA and email parsing, its supported standards and the conversion result. Keep any controlled repair as a separate test fixture. Avoid promising that every provider, gateway and client supports the same internationalised workflow without testing that route.

DMARC tag names and defined values continue to follow their specified syntax. Do not translate v=DMARC1 or invent a Unicode-specific policy version. For internationalised report destinations or provider-generated settings, follow the applicable URI, DNS and provider rules rather than assuming that mailbox display support permits arbitrary Unicode in every configuration field.

Separate visual impersonation from alignment

A displayed name can resemble a familiar brand while representing a different underlying domain. IDNA conversion explains domain representation; it is not a judgment that the name is legitimate or that two visually similar names are the same identity.

When reviewing suspicious mail, compare the actual domains and their DNS forms with the approved identity register. Keep display-name and lookalike-domain findings alongside, but separate from, the DMARC result. The authentication-versus-authorisation guide explains why a valid domain result cannot approve message intent.

Keep a reproducible identity record

For a real investigation, retain the displayed sender, raw From field, extracted Unicode domain, validated A-label conversion, policy lookup names, SPF and DKIM identities, trusted receiver results and transport evidence. Redact personal data without destroying the domain representation needed to reproduce the decision.

A useful result can say that two representations mapped to the same domain, or that they were distinct, and show how that affected the actual evaluation. Preserve the conversion and the original message so another reviewer can reach the same conclusion.

Record the conversion without inventing authentication

Written case IDN-01 retains From info@bücher.example and its corresponding domain xn--bcher-kva.example. The mail administrator records the applicable IDNA library/version and preserves the raw message separately. SPF, DKIM verification, SMTPUTF8 negotiation and discovered policy remain unknown in this case. The next step is to collect those observations, not mark the message authenticated because the two domain representations agree. No message or DNS query was executed.

[ Takeaway ]

Representation conversion is not a rewrite of signed message content.

[ By role ]

What this means for you

Choose your role for practical next steps.

IT / Mail Admin

Compare extracted domain forms, not screenshots.

A reproducible conversion helps distinguish equivalent representations from a real identity mismatch.

Next steps

  1. Preserve original message bytes.
  2. Record validated U-label and A-label forms.
  3. Test the actual SMTPUTF8 and authentication path.

CISO / SecOps

Keep visual risk separate from domain validation.

Valid encoding and passing authentication do not certify a displayed brand or message purpose.

Next steps

  1. Review actual domain identities.
  2. Retain parser and conversion context.
  3. Include internationalised display cases in authorised testing.

Founder / Non-technical

Support the intended identity consistently.

Internationalised branding needs a working mail path and clear recipient presentation, not just a display setting.

Next steps

  1. Identify required scripts and address forms.
  2. Include provider and client compatibility in testing.
  3. Keep approved domain forms in the identity register.

[ FAQs ]

Frequently asked questions

What is the difference between a U-label and an A-label?

They are Unicode and ASCII representations defined by IDNA for internationalised domain labels. Validate the conversion using the applicable rules.

Which form does current DMARC use after extracting a Unicode Author Domain?

RFC 9989 requires a U-label Author Domain to be converted to A-label form for further processing.

Does Punycode apply to the whole email address?

No. Domain-label encoding is separate from the mailbox local part. A Unicode local part is handled through internationalised email rules.

Does an A-label domain alone require SMTPUTF8?

No. Writing the domain in ASCII A-label form does not by itself require SMTPUTF8, although other address or header content may.

Should I rewrite the raw From header before checking DKIM?

Preserve the original message. A domain representation used in evaluation is separate from modifying signed header content.

Can two visually similar internationalised names be treated as aligned?

Not on appearance alone. Apply the actual domain conversion, authentication and alignment rules. Visual resemblance is a separate security concern.

Do internationalised domains use a translated DMARC version or policy value?

No. Use the protocol's defined syntax, including v=DMARC1, and apply the relevant rules to each field rather than inventing translated tags.

What should an identity investigation preserve?

Keep display and raw forms, validated domain conversion, query names, actual authenticated identities, receiver results and relevant transport/parser versions.

[ Sources ]

Sources

  1. RFC 9989

    Sections §4.4 · §5.3.1 · §5.3.5 · §7.4 · §11.7

  2. RFC 5890

    Sections §2.3.2.1

  3. RFC 6531

    Sections §3.2 · §3.3

  4. RFC 6532

    Sections §3

  5. RFC 6376

    Sections §3.4 · §3.6.1 · §3.7 · §6.1.2 · §8.3

[ Glossary ]

Glossary

U-label
A valid Unicode representation of an internationalised domain label.
A-label
The validated ASCII IDNA representation of an internationalised domain label.
SMTPUTF8
An SMTP extension for transporting internationalised email addresses and headers.
Alignment
The protocol-defined domain relationship, not visual similarity.

Enterprise provenance

Why top financial institutions and governments retired their email security dashboards.

Traditional email authentication platforms sell you a dashboard and ask your engineers to do the labour. In high-stakes environments, manual oversight means delayed enforcement, configuration drift and unmonitored shadow domains.

EmailAuth is the agentic email provenance platform. Our agents enforce p=reject, keep third-party senders aligned and move your domains to full RFC 9989 compliance, the new DMARC standard,, with human-in-the-loop auditability.

Governments Large enterprises Tier-1 financial institutions

SaaS · Dedicated tenancy · Air-gapped on-prem · BYOK encryption

Knowledge Base