Five thousand four hundred eighty-one malicious messages rode a real Microsoft 365 incident ID into 1,184 organizations in three days. According to Guardz, operators forged service-health style recovery mail around Exchange Online incident EX1550543, pushed recipients toward a credential-harvesting page on reallifeaccounting.com, and asked staff to type secrets into an attacker-controlled form. Microsoft Threat Intelligence characterized the activity as consistent with credential phishing, Microsoft products blocked the campaign in observed telemetry, and public reporting has not detailed confirmed account takeovers. Phishing-proof device-bound authentication prevents this class of Microsoft 365 outage credential phishing because there is no reusable password left for a blog-shell form to post to a local handler.

What public telemetry shows about the EX1550543 wave

Guardz telemetry counts ten distinct email waves across a three-day window, with each batch completing in 3 to 22 minutes. Earlier traffic on 9 and 10 November 2026 already used domains including newbooker.com and lesterscompany.com. The heavy volume arrived on 13 November 2026 while recovery talk from the 12 November disruption was still live. On the peak day, Amazon SES carried the mail from a short set of aged sender domains that included moments-notice.com along with planetmaison.com and newbooker.com, and each of those domains presented valid SPF and DKIM.

Microsoft’s own EX1550543 event was a network capacity upgrade glitch, not a cyberattack. Outlook, Teams, SharePoint, and Copilot felt the disruption. Operators did not invent that crisis. They forged notices that named the real code and told staff held messages would vanish unless someone clicked a restore-style link. Guardz describes lure body text such as “Mail delivery to your mailbox failed during the Microsoft 365 disruption on 12 November 2026” and “Action required: (4) messages held after the outage.” Links moved through SendGrid-style tracking, then redirected to reallifeaccounting.com/themelist0.html behind Cloudflare.

Landing pages wore generic mindfulness-blog shells. Per-victim tracking beacons and unique session IDs fired before the form rendered. Submitted credentials posted to a local index.php handler on the attacker side. That is password capture on a wrong-origin page. Public analysis does not show a reverse proxy in front of the real Microsoft login, live OTP relay, or session-cookie theft. Microsoft Threat Intelligence has not verified stolen passwords or MFA details. Named victim organizations, kit hashes, and threat-actor brands are not in the public record either. For the full attack-chain narrative, including wave timing and lure copy, see the companion on legacymfa.sucks.

Why outage-themed forms still beat password-only sign-in

Staff already expected recovery mail after a real service-health event. A subject line that mirrors an Outage-plus-incident-code pattern matches that expectation. Valid SPF and DKIM on the SES path only authenticated the attacker-controlled sending domains. They never proved Microsoft authored the brand. Filters that stop many messages still leave a residual path when a user who does click lands on a form that accepts the same secret the real login would accept.

The transferable factor here is the password itself. Anyone who types it into reallifeaccounting.com hands the operator a reusable secret. If that secret still opens the corporate identity provider, the harvest form has done its job before any second factor appears. Recommended cleanup after a suspected submission is familiar: revoke active sessions, force a password reset, and audit sign-in telemetry. Those steps clean up after exposure. They do not change the fact that a typed secret was enough for the kit to collect something useful.

Control in path What the wave used Gap for the user
Outage context Real EX1550543 ID and timing Recovery mail already expected
Mail authentication SES with valid SPF/DKIM Pipe auth, not Microsoft brand
Landing page reallifeaccounting.com form Password typed to wrong origin
Delivery filters Microsoft products blocked many Blocks mail, not lure design

Industrial pacing matters. Ten waves in three days, each finished in minutes, is volume hunting, not a single crafted spear message. Helpdesks still field “I got the outage restore mail” calls for days after a genuine EX-number hits status pages. That pressure is why operators mine real incident IDs instead of inventing fake outages. Password-only or password-plus-phishable second factors leave the same surface the form was built to scrape.

How device-bound phishing-proof MFA empties the harvest form

Device-bound credentials replace the reusable password with a private key that never leaves hardware on an enrolled device. The identity provider holds only the matching public key. A mindfulness-blog shell can still render a fake form. It cannot obtain a signature that proves possession of that private key for the legitimate enterprise origin. There is nothing durable for index.php to store and replay against the real sign-in path.

Public reporting does not describe a reverse-proxy kit for this wave. The documented path is simpler: capture whatever the user types. When the only factor that finishes authentication is a hardware-bound signature demanded by the correct origin, the typed field on reallifeaccounting.com is worthless for completing workforce login. That is Prevention, Not Detection: the remote harvest never yields a usable authenticator, so there is no session for the operator to steal through this kit.

Stage in this wave What attackers sought Device-bound outcome
Initial access Password on harvest form No reusable secret to post
After a real login Not shown in sources Separate endpoint problem
Enrollment or recovery Not the documented path Policy still must stay tight

Public sources have not confirmed account takeovers from this wave. Prevention still applies to the mechanism the kit used. If a password never opens the door, urgent restore copy loses its payoff even when the incident ID is authentic. Session revoke and password rotation remain correct incident response when someone may have submitted a legacy secret. They are cleanup for residual password estates, not a substitute for removing the harvest surface.

Guardz and Microsoft reporting do not describe multi-app SSO token replay as the initial path. The failure mode is earlier: a transferable password on a wrong-origin page. Shrinking blast radius after a legitimate session exists is still good hygiene. It is not what stops this form.

Passkeys harden the login ceremony at the right origin

Passkeys and FIDO2-style WebAuthn put a key pair on the device, keep the private key in hardware, and produce an origin-bound signature for the real identity provider host. That design is origin-bound at authentication. A generic enterprise IdP that issues only hardware-bound signatures will reject a ceremony that never reached its legitimate origin. The EX1550543 harvest page is the wrong origin by definition, so a pure passkey login ceremony does not complete there.

origin-bound login is stronger than OTP, SMS, or push on the strength ladder. It is still narrower than full-lifecycle phishing-proof coverage. Passkeys harden the act of signing in. They do not automatically rewrite every enrollment, device onboarding, or recovery path an organization still runs with email codes or helpdesk-issued secrets. For this campaign the documented ask was a password on a fake form, so origin-bound login already cuts the primary collection path. Teams that stop at passkeys for sign-in alone should still inventory how new devices get trusted.

Full-lifecycle phishing-proof coverage beyond a pure harvest wave

This wave’s public record does not show helpdesk re-enrollment, Temporary Access Pass coaching, or attacker device registration. Enrollment and recovery were not the documented initial-access path. The comparison still matters for defenders who will face the next outage ID tomorrow.

Phishing-proof MFA 2.0 keeps phishable factors out of registration, device onboarding, authorisation, authentication, and decommissioning. Adding a device requires an already-enrolled device rather than a code someone can read aloud or paste into a spoofed page. That full-lifecycle rule is what lifts coverage from origin-bound login to phishing-proof identity. A pure password harvest dies at authentication alone. Coached recovery and fake enrollment die only when those stages lose transferable secrets too. The enrolment model is where organizations close the gap passkeys leave if recovery still issues phishable material.

Operators will mine the next status-page glitch the same way. Filters will catch many waves. The messages that land still fail when the corporate identity stack refuses every factor a wrong-origin page can collect.

Key takeaways for defenders

FAQ

Would phishing-proof MFA have stopped the EX1550543 password harvest form?

Yes. Phishing-proof device-bound MFA would have stopped the EX1550543-themed harvest form from collecting anything that finishes Microsoft 365 workforce sign-in, because authentication would demand a hardware-bound signature for the legitimate enterprise origin rather than a reusable password. According to Guardz, the kit posted submitted credentials to a local handler on reallifeaccounting.com. A private key that never leaves the enrolled device cannot be typed into that page or replayed from index.php.

Did this Microsoft 365 outage phishing wave bypass MFA?

Public reporting has not shown MFA bypass, OTP capture, or session-cookie theft for the EX1550543 outage phishing wave. Microsoft Threat Intelligence described the activity as consistent with credential phishing and has not verified stolen passwords or MFA details. The documented collection path is a wrong-origin password form, not a completed second-factor relay against a live Microsoft login.

Are passkeys enough to stop outage-themed M365 credential phishing?

Passkeys are origin-bound at login and would block a pure wrong-origin password ceremony like the EX1550543 reallifeaccounting.com form when the identity provider requires a hardware-bound, origin-bound signature. Passkeys alone do not guarantee phishing-proof coverage across enrollment, device onboarding, and recovery if those stages still issue email codes, SMS secrets, or helpdesk-reset passwords. Full-lifecycle phishing-proof MFA closes those stages as well.

What should security teams change after an EX-number lure wave?

Security teams should remove transferable passwords from interactive Microsoft 365 sign-in, require device-bound signatures at the real enterprise origin, and audit enrollment and recovery for any remaining phishable codes after an EX-number lure wave such as EX1550543. Delivery filtering remains necessary. It does not replace fixing the factor the harvest page was built to steal. If submissions are suspected, revoke sessions, force resets, and review sign-in logs while the authentication upgrade lands.

Why did valid SPF and DKIM not make the EX1550543 restore mail safe?

Valid SPF and DKIM on the EX1550543 restore mail only authenticated Amazon SES sending for attacker-controlled domains such as moments-notice.com and newbooker.com, not Microsoft as the brand. Recipients saw authenticated mail about a real outage ID, which is how outage-themed credential phishing defeats the shortcut that “authenticated equals safe.” Brand authenticity and factor strength are separate controls from path authentication.