A reverse proxy sat in front of the real Google sign-in page and watched workforce users finish every second factor in real time. Operation CloudFire, reported around 4 November 2026, steered Google Calendar invites and later Drive shares through Google redirectors into that live relay. Victims entered passwords and completed SMS codes, authenticator codes, Tap-Yes number matching, or phone recovery while the Nyasher proxy captured the resulting Google Workspace session cookies and tokens. According to BleepingComputer, covering Material Security's work, those hijacked sessions reached mail, Drive, and the ability to grant OAuth apps in the victim's name. Material blocked 68,676 campaign messages, analyzed 2,112 unique malicious URLs, and took down 897 domains after roughly a month of active disruption.
Public reporting does not name victim organizations or publish exact compromised session counts. The scale of the lure machine still says plenty: 8,570 unique compromised sender addresses across 6,354 root domains showed up in the same window.
How Calendar invites reached a live Google sign-in proxy
Operators started in mid-August 2026 with Google Calendar invites that looked ordinary enough to open. Themes covered voicemail notices, RFPs, summons language, and payment threads. Many invites arrived from previously compromised addresses people already trusted. Links rode Google /url redirectors and related Google-hosted paths, with share.google and Drive document lures appearing as the campaign matured. One Intuit-flavored variant showed up in the mix as well.
Those hops dumped targets onto JavaScript-built capture pages fronted by Cloudflare Turnstile. There was no durable static phishing site sitting around for reputation scanners to burn early. The important piece sat behind the checks: a Nyasher reverse-proxy framework placed in front of genuine Google sign-in. Nyasher had been documented publicly since February after a hardcoded key surfaced in its JavaScript. In this wave it did what reverse proxies are built to do. It presented the real login experience and stayed in the middle of the conversation.
Material Security tracked the traffic from the start and later said detections caught over 99% of malicious messages in its customer base, with a 0.01% reclassification rate. Late September 2026 is when Material stood up Operation CloudFire and began feeding gateway and backend URLs to Cloudflare abuse for takedowns. Over about a month of that pressure, the counters climbed to 68,676 blocked messages, 2,112 analyzed URLs, and 897 domains removed. Lures then shifted harder toward Google Drive file and document shares, and activity appeared to disperse as infrastructure burned.
One infrastructure detail underlines how fast the operators moved. According to Material Security via BleepingComputer, a wildcard certificate was issued 46 minutes before the domain it covered was even registered.
Why SMS, Tap-Yes, and codes still lost the session
The proxy did not need to break cryptography. It needed transferable factors and a patient operator. Victims typed passwords into what looked like Google. Real second-factor prompts fired on real channels. The server relayed SMS codes, authenticator codes, Google's Tap-Yes number matching, and phone-based account recovery to Google as each response arrived. None of those controls stopped the session from being captured.
BleepingComputer's summary of Material's findings puts the human moment cleanly: the victim enters a password, receives a real prompt, and approves it, all while the operator watches the session in real time. That is the entire credential ceremony completed under hostile observation. The prize was not a password file. The prize was authenticated Workspace cookies and tokens issued after MFA succeeded on the proxied path.
| Factor in the path | What the proxy did | Result for the operator |
|---|---|---|
| Password | Relayed live to Google | Login continued |
| SMS or authenticator code | Relayed as typed | Second factor cleared |
| Tap-Yes number matching | Relayed approval | Prompt satisfied |
| Phone account recovery | Relayed recovery step | Account still unlocked |
| Issued Workspace session | Captured cookies/tokens | Mail, Drive, OAuth ready |
Standard Google second factors failed here because every one of them is something a human can type, read aloud, or tap while a middlebox copies the outcome. The login origin the user thought they were binding to was not the origin that kept the session. Public sources do not detail passkey or other origin-bound MFA outcomes for this campaign, and they do not claim those controls were in play for the hit accounts.
An unfinished Microsoft 365 phishing branch sat in the same operator kit, including stylesheet paths, a second harness, and a Microsoft sign-in title string. Those paths did not resolve at reporting time. The live damage stayed on Google Workspace identities.
What hijacked Workspace sessions unlocked next
Once the cookies left the proxy, operators held live Google Workspace sessions rather than a reusable password alone. Material's framing is direct: when the outcome is a hijacked session rather than a stolen password, the evidence shows up in Workspace audit logs. Defenders hunting only for impossible-travel password events miss the plot.
A live Google session reached mail and Drive. It also carried the ability to grant OAuth access in the victim's name, so the blast radius extended well past a single inbox. From there the work is classic post-auth containment: revoke sessions fast, review and kill unexpected OAuth grants, reset credentials, and treat every integrated Workspace surface as exposed until proven otherwise. Shortening cookie lifetime after the fact does not unwind a session the attacker already holds. It only narrows the window if they have not already enrolled lasting access paths.
That sequence is why calendar-themed adversary-in-the-middle campaigns keep landing in enterprise inboxes. An adversary-in-the-middle attack places a live reverse proxy between the user and the real identity provider so password and second-factor traffic can be relayed and the issued session collected. The lure borrows Google's own trust graph. The proxy borrows Google's real login chrome. The second factors still move across the wire as copyable proofs. The session that follows is a skeleton key for the user's Workspace world until someone tears it down.
Detection and takedown still mattered. Nearly nine hundred domains burned and tens of thousands of messages blocked cut the campaign's runway. Prevention at the login ceremony is a different layer. Device-bound, origin-bound authentication that refuses transferable codes and push approvals removes the relay surface this kit lived on. Enrollment and recovery paths that cannot be completed as typed secrets on a hostile page close the same failure class when phone recovery would otherwise hand the attacker a fresh factor.
For the control side of this Google Workspace session-capture path, see how phishing-proof MFA stops Google Workspace AiTM session capture.
FAQ
How did Operation CloudFire steal Google Workspace access?
Operation CloudFire stole Google Workspace access by placing a Nyasher reverse proxy in front of real Google sign-in and driving victims there with Calendar invites and Drive shares that used Google redirectors. Users completed passwords plus live MFA while the proxy captured the authenticated session cookies and tokens. According to Material Security reporting via BleepingComputer, those sessions opened mail, Drive, and OAuth grants in the victim's name.
Did MFA stop the CloudFire Google Calendar campaign?
MFA did not stop the CloudFire Google Calendar campaign for accounts that completed standard Google second factors on the proxied login. SMS codes, authenticator codes, Tap-Yes number matching, and phone recovery were all relayed in real time, so the second factor cleared and the session still went to the operator. The failure was transferable MFA completed on a hostile middlebox, not a missing password.
What should defenders look for after a CloudFire-style Workspace hijack?
Defenders should look in Google Workspace audit logs for hijacked-session evidence rather than only for stolen-password events after a CloudFire-style attack. Priority actions are immediate session revocation, review of OAuth apps granted in the user name, credential resets, and containment across mail and Drive. Material Security and BleepingComputer both stress that the useful forensic signal lives in session and grant activity once MFA has already succeeded under the proxy.
How large was the Operation CloudFire lure infrastructure?
Operation CloudFire's tracked lure infrastructure included 68,676 campaign messages blocked, 2,112 unique malicious URLs analyzed, 897 domains taken down, and 8,570 unique compromised sender addresses across 6,354 root domains, according to Material Security figures reported with the November 2026 disclosure. Public sources do not publish exact counts of stolen sessions or named victim organizations, so those message and domain numbers are the clearest scale markers available.
Was Microsoft 365 part of the live CloudFire attacks?
Microsoft 365 was not part of the live CloudFire attacks described in the disclosure. Operators shipped unfinished Microsoft 365 phishing code beside the live Google paths, but those Microsoft routes did not resolve when researchers reported the campaign. The confirmed session hijacks ran against Google Workspace sign-in and Workspace apps.