Skip to content
Now

Tech · Microsoft 365 outageContinuing story · report 3 of 5

Microsoft outage delays medical scans and a final job interview as recovery remains partial

Australians describe postponed tests, interrupted care work and a final job interview delayed another week. Microsoft reports better stability, but sustained worldwide recovery remains unconfirmed.

Written 6 min read · 167 sources

What changed since our last report

1 hour later

Now

Australians describe postponed tests, interrupted care work and a final job interview delayed another week. Microsoft reports better stability, but sustained worldwide recovery remains unconfirmed.

What moved

  • The Irish Times reported that servers restored repeatedly overnight had crashed again. The earlier stability account had been overtaken by renewed failures.
  • Microsoft reported better service stability and capacity after traffic optimization, while broadening the measures across affected regions in pursuit of persistent recovery—not declaring it achieved.

Maria Derosa needed medical scans after a car accident. But neither she nor her Melbourne clinic could retrieve the referrals through email or fax. The tests were cancelled or delayed, she told ABC News—a broken communications link standing between a patient and the checks she needed.⁠4

That is the practical cost behind Microsoft’s cautious recovery language. The company has reported improved server stability during its worldwide Microsoft 365 outage, but has not confirmed full restoration. Its public incident notice, authored at 07:29 UTC on November 13 and retrieved again at 08:39 UTC, still classified the incident as “Service degradation.”⁠1⁠167

In our November 13 report at 08:16 UTC, we described repeated recovery failures and disruption to government correspondence, public-service transactions and financial-sector work. The additional accounts make those consequences more specific: delayed tests, interrupted care-related work, lost study support and a final job interview pushed back another week. They add evidence of what the outage costs people, not evidence that recovery is complete.⁠4⁠6⁠20

In Melbourne, Manny Flores could not get into Teams for that interview. He had been unemployed for almost a year and reached the final stage; the meeting was rescheduled for the following week. For him, a failed connection meant another wait for an answer about work. In Adelaide, Peter Roche said authentication problems prevented him accessing accounting software and processing invoices. Orders were not arriving and invoices were not going out. He described losing thousands of dollars—his account of his business, not an audited estimate of wider losses.⁠4

Occupational therapist Lauren Van Ingen, also in Melbourne, told ABC that being locked out of Outlook stopped work helping people with motor neurone disease, Huntington’s disease and other neurological conditions obtain equipment to live safely at home. Derosa’s account separately describes an interruption to obtaining medical tests; it does not establish a resulting injury or medical outcome.⁠4

Earlier assurances that particular emergency or clinical services remained available must be read at their actual scale. Queensland’s government said emergency health services and patient care had not been compromised. That was an attributed assurance about its services, not proof that every patient elsewhere could obtain a referral or keep an appointment. The Melbourne accounts do not need to contradict that assurance to matter.⁠141⁠4

Education had its own dependencies. Melbourne student Emilie Yu said she lost access to teachers’ help and work while preparing for two remaining VCE language exams. Susan Wallace in Lismore described losing access to Teams virtual classrooms used at TAFE NSW and elsewhere. Neither account supplies a total of cancelled classes or affected students.⁠4

Why recovery has not held

The incident began around noon on the US East Coast on November 12 and continued into November 13 across Asia-Pacific and Europe. American descriptions of a November 12 outage and Japanese accounts of an early-November-13 disruption concern the same continuing incident across time zones, not an established second failure.⁠1⁠19

Microsoft’s explanation is that unexpectedly unavailable servers reduced usable capacity, leaving the remaining infrastructure carrying too much work across multiple regions. The company distinguished that concentration of load from an increase in customer demand. Restarting servers could restore capacity temporarily without making it resilient: some stopped responding again when traffic returned.⁠2⁠20⁠6

Progress, setbacks and cautious improvement

  1. November 12, 23:47 UTC

    TV Asahi reported limited recovery, with Microsoft saying some restarted servers were stable so far.

    19
  2. November 13, 06:27 UTC

    The Irish Times reported that servers restored repeatedly overnight had crashed again. The earlier stability account had been overtaken by renewed failures.

    6
  3. November 13, 07:44 UTC

    Microsoft reported better service stability and capacity after traffic optimization, while broadening the measures across affected regions in pursuit of persistent recovery—not declaring it achieved.

    1156

The investigation included automated health-check failures and an overwhelmed process responsible for establishing secure connections. A possible problem in a component processing server health remained a hypothesis. These details help explain how servers could stop processing requests; they do not identify what first made them unavailable.⁠4⁠10⁠158

Engineers paused deployments, redistributed traffic, restarted servers and tested configuration changes and alternative health checks. The 07:29 public notice described positive stability results after an upstream authentication/routing configuration-component patch was applied to part of the infrastructure. Expanding a promising repair is meaningful progress, but a repair involving authentication or routing does not prove that a defective update caused the original outage.⁠8⁠23⁠167

That notice promised another update by 08:30 UTC. It was a target for information, not a restoration deadline; the copy retrieved afterward still carried its earlier authored timestamp. No full-restoration time was established.⁠167

Microsoft executive Charles Lamanna said there was “no indication” of malicious or cyber activity. The qualification matters: an absence of a present indication is not a completed forensic exclusion of every security cause. Microsoft has promised a post-incident explanation of the root cause and defensive repairs once the incident is mitigated, without a publication date.⁠8⁠167

One dependency, many different failures

The public notice lists 20 product and service entries, not 20 wholly unavailable applications. Categories overlap, and the central distinction is functional: features dependent on Exchange Online and email could fail while other parts of the same product remained available.⁠167

Exchange users could lose mailbox access, miss their newest messages or face delayed delivery. Most core Teams calls, meetings, group chats and channel messages remained available, while calendar scheduling, invitations and some event or channel actions could fail. SharePoint users could still access stored files directly. Those boundaries explain how one person could keep working while another lost an essential workflow.⁠3⁠167

The effects also reached education tools, device enrollment, task management, Copilot access to work information, and security and compliance functions. Microsoft reported possible failures involving Defender’s Safe Links, Safe Attachments, quarantine releases and email remediation, alongside Purview labeling, protected-content access and legal discovery. These are protection-feature and access failures—not evidence that attackers breached customers.⁠167

Microsoft-connected AI tools had separate unresolved problems. OpenAI said new Microsoft connector connections might break and tools might not work as expected, without identifying the individual tools. Anthropic’s latest documented update, at 22:23 UTC on November 12, said its connector’s mail, calendar and search tools were still failing for most requests and Claude for Outlook remained affected. No confirmed full restoration was established; neither notice meant all of ChatGPT or Claude was down.⁠6⁠11⁠16

The reach exceeds the available count

Downdetector reports reached 74 countries and territories in about an hour. The BBC described a worldwide Outlook complaint peak of nearly 140,000; RTÉ reported a US-specific peak above 130,000. Those figures have different geographic scopes and are complaint reports, not audited unique people. They cannot be added together into an outage toll.⁠1⁠2⁠16

The consequences were unmistakably international. Los Angeles-area law-enforcement agencies lost email or other essential communications. New Zealand’s foreign ministry reported disrupted email reception, including at diplomatic offices worldwide. Japan’s government switched to physical documents, while Hong Kong employees at Morgan Stanley and Goldman Sachs described losing email or Teams through their lunch break. None of that establishes that entire government or banking operations stopped.⁠7⁠5⁠1

Travel accounts need the same precision. Sydney resident Livinia Drummond told ABC that a colleague’s flight to Papua New Guinea was delayed while printed copies of documents sent through Outlook were brought to the aircraft. That second-hand account did not establish the carrier or delay length. Separately, IndiGo, Akasa Air and CemAir reported email or system difficulties without reporting flight disruptions; Air India said operations remained unaffected. This was not evidence of worldwide flight groundings.⁠4⁠1

One business consequence has a fixed new date. Workhorse Group postponed both its third-quarter earnings release and results-and-business-outlook call because of the disruption. The call moved to November 16 at 4:30 p.m. Eastern; the release will precede it. That is a documented delay, unlike any proposed worldwide financial-loss total.⁠16⁠146⁠153

Known

  • Microsoft has reported recovery progress, while named people describe interrupted work, care and study. Full restoration remains unconfirmed. 14167

Unknown

  • The initiating cause, audited totals of affected users and organizations, aggregate losses and full connector-restoration times remain unresolved.

Next

  • Whether restored capacity stays responsive as traffic returns—and whether Microsoft’s promised incident report explains why repeated recovery attempts failed.

An inbox working briefly is not the finish line. The test is whether the systems stay available long enough for the referral to arrive, the invoice to leave and the interview to happen.

Timeline

Day 4 · 5 reports

Story timeline
  1. Microsoft's stated end of active customer impact.

  2. Microsoft says service is restored to the remaining affected customers and regions and keeps monitoring delayed mail.

  3. Preliminary report published; Lamanna apologises on X.

Sources123

Of the sources cited here, Claude published first, 3 days before this report aired. 89 more outlets followed.

33 outlets with no date:Workhorse Group, Inc.cloud.microsoftreddit.comInvesting.com भारतnacionale.comRed Hot Singapore

Continuing story · report 3 of 5

Every source, by outlet

167 sources from 123 outlets· Most cited first

As aired
  1. Microsoft's outage has delayed medical scans and a final job interview. The company reports better server stability, but it still classifies the incident as service degradation.
  2. On November 13, we reported repeated recovery failures and disruption to government correspondence and financial-sector work. The additional accounts show what those failures cost people. Can the repaired systems stay responsive as people return to work?
  3. Manny Flores couldn't get into Teams for his final job interview in Melbourne. He'd been unemployed for almost a year. The interview was moved to the following week, extending his wait for an answer about work.
  4. In Adelaide, Peter Roche said authentication problems blocked his accounting software and stopped him processing invoices. Orders weren't arriving and invoices weren't going out. He described losing thousands of dollars.
  5. Melbourne occupational therapist Lauren Van Ingen said she was locked out of Outlook. That stopped work helping people with motor neurone disease, Huntington's disease and other neurological conditions obtain equipment to live safely at home.
  6. Maria Derosa, also in Melbourne, said scans after a car accident were cancelled or delayed. She and the clinic couldn't retrieve referrals through email or fax. She described being unable to get the tests she needed to check that she was okay.
  7. Emilie Yu was preparing for two remaining VCE language exams in Melbourne when she lost access to teachers' help and work. In Lismore, Susan Wallace described losing Teams virtual classrooms used at TAFE NSW and elsewhere.
  8. Those accounts add specific interruptions to education and care. Earlier assurances about available clinical services applied to particular organizations and functions. They couldn't guarantee that every patient could obtain a referral or keep an appointment.
  9. The disruption crossed working days and continents. Public agencies lost communications in Los Angeles. Bank employees couldn't access email or Teams in Hong Kong. Government systems were affected in Singapore, while Melbourne residents described interrupted work, care and study.
  10. The incident began around noon on the US East Coast on November 12 and continued into November 13. The different dates in American and Japanese accounts describe the same continuing outage across time zones. By Friday morning, it had already lasted more than twelve hours.
  11. Outlook complaints peaked at nearly 140,000 worldwide. The US peak exceeded 130,000, with more than 25,000 still being logged that evening. These are complaint reports with different scopes, rather than a count of unique people.
  12. Complaints reached 74 countries and territories within about an hour. Nearly three in ten reports during an early reporting window came from outside the United States. Users in Denmark were among those affected; the number affected in Finland wasn't known.
  13. In Singapore, Outlook reports rose from about 730 to 1,130 in an hour. Microsoft 365 reports rose from 434 to 566 during a separate morning window. Different products and reporting windows can't be combined into one global outage curve.
  14. Microsoft says more than 430 million people use its Microsoft 365 apps, and more than 90 percent of Fortune 500 companies use Copilot. Those adoption figures explain the dependency. Nobody has supplied an audited total of affected users or a worldwide financial-loss figure.
  15. Subscribe to this channel to follow whether Microsoft's recovery holds and how essential services recover.
  16. Some restarted servers initially looked stable. Later, restored servers stopped responding again. A gradual-recovery announcement in Australia was followed about an hour later by renewed unresponsiveness. An inbox briefly working again couldn't guarantee that the repair would hold.
  17. Microsoft says servers unexpectedly became unavailable, leaving the remaining infrastructure carrying too much work across multiple regions. The company distinguished that loss of capacity from a surge in customer demand.
  18. Losing servers concentrated requests on the machines still operating. Some couldn't process requests reliably or pass their health checks. Restarting them restored capacity temporarily, but returning traffic sometimes made them unresponsive again.
  19. The investigation included automated health-check failures and an overwhelmed process responsible for establishing secure connections. A suspected problem in a component processing server health remains a hypothesis. Microsoft hasn't identified what first made the servers unavailable.
  20. Microsoft's early changes weren't taking as expected. Engineers continued recovering frontend servers and adjusting traffic. Later updates described positive server-stability results after a patch, followed by better stability and capacity from traffic optimization.
  21. Engineers paused deployments, redistributed traffic and restarted servers. They tested alternative health checks and configuration changes, then checked stability before bringing more traffic back. A repair involving authentication or routing doesn't tell us that a defective update caused the original failure.
  22. The patch brought positive stability results on part of Microsoft's infrastructure. Engineers were expanding it. The later traffic changes were also being broadened across affected regions to achieve persistent recovery. Microsoft still hadn't announced full restoration.
  23. The public notice promised another update by 08:30 UTC. That was a target for information, rather than a restoration deadline. The notice retrieved afterward still carried its earlier authored timestamp and service-degradation status.
  24. Microsoft executive Charles Lamanna said there was no indication of malicious or cyber activity. Microsoft's public notice used similarly limited language. The company hasn't presented a completed forensic exclusion of every possible security cause.
  25. Australian Home Affairs Minister Tony Burke compared the failure to an update going wrong while relaying Microsoft's assessment. That comparison didn't identify the initiating cause. Microsoft's own investigation into the server failures continued.
  26. Microsoft's public notice lists twenty product and service entries. An earlier local account counted nineteen. The snapshots and overlapping categories don't show that exactly one more standalone application failed. The notice describes email-dependent features failing while other functions remain available.
  27. Most core Teams calls, meetings, group chats and channel messages remained available. Calendar scheduling and invitations could fail, along with some event and channel actions. Standard channels worked, and SharePoint users could still access stored files directly.
  28. Exchange Online was central to the disruption. People could lose mailbox access through web, desktop and mobile connections, miss their newest messages or face delayed delivery. Some Microsoft 365 administration pages failed to load, and the apps could lose their recent-files view.
  29. Education assignments and reading and speaking tools could be impaired. Copilot could fail to respond or access work email, calendars and files. Device enrollment and compliance checks, Planner tasks and To Do operations could also become unreliable.
  30. Microsoft Graph requests for email, calendars and contacts could time out, interfering with Teams management and notifications. Some Power Apps and Field Service mobile sign-ins failed intermittently. Universal Print could fail to submit jobs, while some Windows 365 customers using Purview Customer Key could lose connectivity.
  31. For Power Apps and Field Service, web access could provide an alternative. Already signed-in users could potentially avoid disruption by staying signed in and avoiding data clearing or reinstallation. That advice applied to those applications, with no universal fix for the outage.
  32. Security and compliance tools also lost functions. Defender for Office 365 could fail on Safe Links, Safe Attachments, quarantine releases and email remediation. Message traces, reporting and protection changes could be delayed. Defender XDR portal pages might not load.
  33. Purview labeling, data-loss prevention, authorization and classification were affected. Protected content, audit searches and legal discovery could fail. Fabric and Power BI could also struggle with sensitivity-labeled files or exports. These were protection-feature failures; they don't demonstrate that attackers breached customers.
  34. Removing a sensitivity label was described as a possible workaround only where an organization's data-protection policies allowed it. The outage didn't make it safe to disregard those policies.
  35. The Los Angeles County Sheriff's Department lost essential communications and directed inquiries to its official website. The LAPD and county district attorney also had disrupted email. Torrance said municipal services would continue. Colorado's Arapahoe Sheriff's Office reported delayed bond payments for prisoners.
  36. Miami Beach, Austin and Jacksonville reported email failures while keeping emergency services available. Baltimore County urged people to telephone departments. Pasadena, Fremont, Burbank and Santa Barbara also warned that email was down.
  37. Japan's government switched to physical documents after email disruption. The Philippines' Supreme Court warned of delayed email responses and transactions. New Zealand's foreign ministry lost email reception, including at diplomatic offices worldwide. Australian intelligence and federal-police organizations were also affected.
  38. Singapore's earlier assessment kept emergency calls to 995 and 999, Singpass and public officers' internal Teams communications available. GovTech warned that some replies could be delayed. Service NSW advised postponing service-centre visits and using online transactions where possible.
  39. The University of Iowa hospital network directed patients to its online portal after email and voicemail disruption. Several UK general practices warned of possible problems and directed urgent queries to the telephone rather than email or web forms.
  40. A travel account showed another email dependency. Sydney resident Livinia Drummond said a colleague's flight to Papua New Guinea was delayed because the captain's documents had arrived through Outlook. Passengers waited on a tarmac bus while someone brought printed copies on foot.
  41. That second-hand account didn't identify the carrier or the delay's length. IndiGo, Akasa Air and CemAir described email or system difficulties without reporting flight disruptions. Air India rerouted email and said operations and communication channels remained unaffected.
  42. IndiGo directed customers to its website or mobile app while incoming email support was impaired. Those different experiences show why an email outage can't be expanded into a claim of worldwide flight groundings.
  43. Hong Kong employees at Morgan Stanley and Goldman Sachs described losing email or Teams from logging on through lunch. Nomura staff also reported problems. In Raleigh, a medical-supply salesperson couldn't send quotes, and legal work was interrupted.
  44. Microsoft-connected AI tools had their own problems. OpenAI said new Microsoft connector connections might break and tools might fail. Anthropic said its connector's mail, calendar and search tools were failing for most requests, with Claude for Outlook also affected.
  45. Those connector notices didn't mean every part of ChatGPT or Claude was down. Anthropic also described incomplete connections and Outlook add-in failures. Neither company had provided a confirmed full restoration of these Microsoft-connected functions.
  46. Banks warned that criminals could exploit the confusion. Standard Chartered Singapore, Bank of Queensland and ING Australia cautioned about fake technical support, requests for personal information or one-time passwords, and attempts to gain remote access to devices.
  47. Workhorse Group postponed both its third-quarter earnings release and its results-and-business-outlook call because of the disruption. The call moved from November 12 to November 16 at 4:30 p.m. Eastern. The release will precede it.
  48. The patch and traffic changes have improved stability, but Microsoft is still extending them across affected regions to achieve persistent recovery. It hasn't given a time for full restoration. Microsoft has promised a post-incident explanation once the incident is mitigated, without setting a publication date.

Aired recently