The IETF ran an Email Address Verification BoF on 17 November with draft working-group charter text on the slides. The proposed flow proves a person controls an address without sending them mail — and says in writing that it does not prove the mailbox can receive it.
On Tuesday 17 November 2026, the IETF's Applications and Real-Time area held a two-hour Birds-of-a-Feather session in San Francisco called Email Address Verification. The agenda is short: a problem statement from Dick Hardt, a talk on origin-bound codes and WebOTP from Tess O'Connor, Eryn Wells and Sam Goto, the two internet-drafts that describe the mechanism, and a closing half hour on charter text. That charter — for a working group that does not yet exist — is on the slides.
The proposal is to verify that someone controls an email address without sending them an email. Read as a signup-flow optimisation, that is mildly interesting. Read against what your CRM column means, it changes the meaning of a field you have been trusting for a decade.
In the draft the browser sits in the middle. The user types an address into your form; the browser requests a signed token from an issuer; the issuer returns a token binding the address to a key the browser holds; your site verifies the signature. No message is delivered, so nothing can bounce, and the token is checked at the moment of signup rather than minutes later when someone opens an inbox.
Who is allowed to assert an address is not your decision, and it is not the site's either. The domain publishes a DNS record at _email-verification. followed by the domain, carrying iss= and the issuer's host. The draft's reasoning for DNS is worth keeping: access to DNS and access to mail are usually separate from access to a website deployment, so someone who can only edit a web property cannot mint verified addresses for that domain.
A second design does the same job for the message-based flow. Sam Goto and Mike West's OTP-Token draft defines a machine-readable, origin-bound field inside the verification email, so the mail client can hand the code to the browser instead of asking the user to copy digits across applications. The BoF charter names both drafts as starting points rather than competing options.
The draft states the limit itself, in language worth quoting because it is the sentence that should end up in your documentation:
"Like social login, this protocol verifies that the user controls an email address — it does not verify that the email address can receive mail."
Hardt & Goto, Email Verification Protocol, draft-hardt-email-verification-02, §1 (August 2026)That is not a drafting gap; it is structural. There is no SMTP transaction in the flow, so no MX conversation happens, no recipient is refused, and no bounce exists to record. Your suppression logic, your bounce-rate dashboard and your mailable-audience count are unchanged by a verified token. What changes is the funnel: friction at the verification step drops, and the population of addresses that pass a check called "verified" grows. If you have been reading verification rate as a proxy for list quality, this protocol quietly invalidates the proxy while making the number look better.
This is not hypothetical work sitting behind an RFC. The draft's implementation section records Gmail as an issuer at origin trial maturity, with EVT issuance, DNS-based discovery and a JWKS endpoint, plus Chrome and Edge origin trials on the browser side. You may see token-based verification in your own funnel data before the document reaches Proposed Standard.
The draft also defines private addresses. A browser can ask for private_email: true and receive a per-site relay address that the issuer routes to the real mailbox. On a later visit it can pass directed_email to get the same address back if it is still linked to that account, or an error if it is not. The requirement is that the private address be unique per user and per relying-party origin, and that it route.
Three operational consequences follow, and none are addressed by the draft:
The charter's out-of-scope list is the most consequential page in the deck. It excludes, in the drafters' own words, "Consent to receive messages at the address." It also excludes "How an application uses a verified address, such as for account recovery or as an authentication factor."
So a verified token is proof of control, and nothing else. It says the person at this browser can reach that mailbox right now. It does not say they asked you to write to them, and it does not replace the record — a double opt-in click, a preference-centre choice, a timestamped consent string — that a privacy review asks you to produce. If any part of your compliance story currently rests on the sentence "the address was verified," a standards document is now on record saying that sentence does not carry the weight.
The authentication exclusions are just as deliberate: no changes to DKIM, SPF or DMARC are in scope, and SMTP and the message format stay as they are apart from one new header field. If you were hoping a verification standard would arrive as a deliverability fix, it is not that. Deliverability stays where it is: your authentication, your sending history, your engagement, your complaint rate.
Among the criteria the charter asks the working group to weigh is "whether the mechanism reveals if an address exists." That is a live question in this industry for unglamorous reasons: existence is currently inferred from SMTP recipient probes, catch-all behaviour and paid verification APIs that sell an answer with varying levels of confidence behind it.
The draft's own error handling points one way — uniform responses that do not distinguish "no such address" from other failures, and mitigations against inferring existence from response timing. A working group that encodes that default would be treating mailbox existence as something a website should not be able to buy from a browser. That is a different position from the one a segment of the list-hygiene market sells, and it is worth knowing which way the reference design leans before you renew a contract premised on the other one.
The Datatracker lists evp as a BOF in the ART area, chaired by Bron Gondwana and Justin Richer, with Andy Newton as area director, and a draft charter, charter-ietf-evp-00-00, marked "Not currently under review." Chartering a working group is the IESG's decision, and no outcome from the Tuesday session had been published when this was written. A BoF is a room agreeing there is a problem; it is not a standard.
Three things still have to happen before any of this touches a send. Mail clients need to implement the message-side half; domains need to publish the delegation record, which puts the decision in the hands of mailbox providers rather than senders; and the browser API itself is being specified elsewhere, in the W3C's WICG origin-bound code work that the charter cites as precedent. The draft expires on 26 February 2027. Senders will not get a vote on adoption, which is exactly why the definition of a verified address is worth reading while it is still a draft.
Take one signup form and write down what its verification step proves. Today the honest answer is that someone read mail at that address, minutes ago. Under this protocol it becomes: an issuer the recipient's domain trusts says the person at this browser controls it. Neither statement is consent, and neither one is a mailable mailbox. Find every downstream system that treats "verified" as either of those, and fix the label before the tokens arrive.
Written by the team at LMR — a senior lifecycle marketing studio working in Iterable, Braze, Klaviyo and the tools you already run.
Start a project ↗