Huntress counted stolen Microsoft 365 sessions at 79 organizations. The count is Huntress's, and a password reset is not the whole cleanup.
Huntress says stolen Microsoft 365 sessions were reused at 79 organizations from July 14 to September 22. Two outlets retell that research. This page prints no addresses.
The short version
Huntress says that from July 14 to September 22, 2026, stolen Microsoft 365 sessions were reused at 79 organizations: at least 497 successful sign-ins and more than 200,000 failures. It reports new devices, replaced sign-in methods, and inbox rules, and says a password reset leaves the devices in place. Two outlets retell the research. This page does not print addresses or steps.
A research count, with two retellings
Huntress published its account on October 22, 2026. It says a campaign running from July 14 to September 22 reused stolen Microsoft 365 sign-in sessions across 79 organizations where Huntress was engaged. The post's own totals are at least 497 successful sign-ins and more than 200,000 failed attempts. Help Net Security's October 23 story uses those totals and the same dates. SecurityOnline's summary does too, and it labels the source as Huntress. There is no fourth document here from Microsoft that adopts the number 79. A vendor tally and a product company's tally would be different. This page has the first.a session is not a bag of leftover state.
What kind of theft the post describes
Huntress says the password was not the thing being guessed at the moment of the successful sign-in. The person had already completed sign-in, including the multifactor step, and the resulting session was then used from somewhere else. The post discusses more than one way a session can be taken. This page does not walk through those ways, name the tooling, or print the network ranges the researchers tracked. The impact that belongs here is the session's value: it is proof the sign-in already happened, so a later use does not have to repeat it. Failed attempts in the hundreds of thousands are Huntress's picture of trying many stolen sessions, not of a single mailbox.what an advisory must name.
Devices and rules are the residue
Across the environments, Huntress says it saw at least 48 devices registered into the directory by the activity, sign-in methods wiped and replaced in 19 organizations, and inbox rules used to keep victims from seeing some of their mail. It describes one case in which a compromised mailbox sent a payroll-related message outward. Help Net Security repeats the hidden-rule point and the hosting detail this page is leaving out. The residue is the part a password reset misses, which is Huntress's explicit warning. A new device and a new sign-in method survive the reset. That is why the research says cleanup has to look at both.Storm-0940 is a different Microsoft account.
Dates and a quiet week
The window Huntress gives is ten weeks, ending September 22. The post describes bursts, a change of infrastructure, and a final session on the last day of the window. Those phases are the researcher's reconstruction. This page keeps the start, the end, and the totals, and it does not republish the phase diagram or the addresses inside it. Help Net Security's value is the plain restatement on October 23, which is why the item is on the wire today even though the research date is October 22. SecurityOnline adds a compact table that still points back at Huntress. Neither outlet is a second investigation with its own logs.the catalog date is a different list.
What would have stopped a sign-in, on Huntress's account
Huntress says conditional access policies, properly applied, can refuse this class of replay, and that in four of its customer organizations those policies blocked attempts from the activity. A control that worked in four tenants is not a control that was on in the other seventy-five. The post is arguing for the control and reporting where it saw the control work. It is not a Microsoft default changing on October 23. This page names the control at that level and does not turn the post into a setup guide.an identifier is not a how-to.
Who is not in the 79
An organization that is not a Huntress customer is not in the count, even if the same activity reached it. The 79 is a denominator of engagements, which is a real denominator and a limited one. A reader who treats it as the size of the campaign worldwide is inventing a census. Huntress does not claim it saw every tenant. It claims it saw these. The payroll example is one incident inside the set, not a finding that every one of the 79 lost a paycheck.Capcom's count is a different breach.
Who is in scope
Microsoft 365 customers trying to tell a stolen session from a stolen password. The 79 organizations, as Huntress's engagements, unnamed here. Administrators who would have to look for unexpected devices and sign-in methods, not only reset a password. People outside Microsoft 365 are out of scope. Microsoft, as the maker of the service, is in scope as the platform and not as the author of this count.PHP's later note is a different product.
What stays on the record
Huntress, October 22, 2026. Window July 14 to September 22, 2026. 79 organizations. At least 497 successful sign-ins. More than 200,000 failed attempts. At least 48 registered devices. Sign-in methods replaced in 19 organizations. Inbox rules. One described payroll message. Password reset does not clear devices or methods, on Huntress's account. Conditional access blocked attempts in four organizations, on that account. Help Net Security, October 23, and SecurityOnline retell the research. No Microsoft bulletin in this source list. No addresses and no steps on this page.
What would move the row
A Microsoft incident report that adopts or corrects the totals would be a later document. A named-victim filing would be a different kind of record. Until then the record is Huntress's research and two retellings. The short answer this page will stand on is this: Huntress says that from July 14 to September 22, 2026, stolen Microsoft 365 sessions were reused at 79 organizations: at least 497 successful sign-ins and more than 200,000 failures. It reports new devices, replaced sign-in methods, and inbox rules, and says a password reset leaves the devices in place. Two outlets retell the research. This page does not print addresses or steps.the advisory is the story.
Sources
The reports this brief is filing. Futureweb did not republish them.
Questions
Is this a Microsoft advisory?
No. The primary document is Huntress's research post, dated October 22, 2026. Help Net Security's story is October 23. This page did not find a Microsoft security bulletin that owns these figures. The figures stay Huntress's.
Did the attacker need the password?
Huntress says the accounts were reached without the attacker needing the password, because the session was taken after the person had finished signing in, including multifactor authentication. A session that already exists is not the same fact as a guessed password. This page does not describe how the session was taken.
What did Huntress see after the sign-in?
It says the activity registered at least 48 devices in the customers' directories, replaced multifactor methods in a subset of the organizations, and created inbox rules that hide mail. In one incident it says a payroll-related message was sent to an outside contact. Those are outcomes in the research, not a procedure.
Does a password reset end it?
Huntress says no. A reset does not, by itself, remove a device the attacker registered or a new sign-in method the attacker added. The research says both have to be removed, and the sessions revoked. That is the researcher's containment advice, stated as a result, not as a command list.
Are the 79 organizations named?
No. Huntress describes them as customer environments it responded in. A count of organizations is not a victim list. SecurityOnline's table attributes the same count, the same dates, and the same sign-in totals to Huntress.