Yesterday afternoon, email stopped for a huge part of the working world, and for many people it was still down this morning. Microsoft says it was not a cyberattack, which is almost beside the point: if your email, calendar and chat vanished for half a day, your business felt the same thing a ransomware victim feels. This week we use the outage as a lesson in resilience, plus a B.C. application portal that has been offline for over a week, an extremely open set of AI-built apps, and the credentials of 6.4 million people sold to criminals. Let's get into it.

🗞️ STORY OF THE WEEK

Microsoft's Outage Shows How Fragile 'Always On' Really Is

Starting yesterday afternoon, Outlook, Exchange Online, Teams and Copilot went down for users in dozens of countries, with Downdetector logging reports from about 74 countries. Government departments and courts in several countries were disrupted, and some AI services that connect to Microsoft were hit too. By Friday morning Microsoft was reporting only "marginal success" restoring service, more than 12 hours in. Servers it brought back were failing again once real traffic returned.

Microsoft has not published a root cause. It says affected servers were failing automated health checks and that a core process on them was becoming overwhelmed, and it was rolling back a recent network change. A senior Microsoft executive said there was no indication of malicious activity, though The Register noted the company had not fully ruled it out. Australia's Home Affairs Minister said Microsoft compared it to the CrowdStrike failure of 2024, where an update went wrong. The company's own admin tools, including Intune, went down too, which meant IT teams could not easily manage their own systems while it was happening.

For a business owner, the cause matters less than the effect. Whether the trigger is a cyberattack, a bad update or a network fault, the result is the same: your main tools are gone and you do not control the fix. This is also not Microsoft's first outage this year. Al Jazeera noted a near ten-hour outage in January and an Outlook lockout in late August. As one analyst put it, the question is: what is the minimum we need to keep running if this happens again?

Why it matters
Most small businesses run email, calendar, files, chat and sign-in all through one provider. If it goes down, can you still reach your customers, your staff and your suppliers? If you rely on email to receive orders or confirm payments, how long could you go without it?

What can you do
Today, make a one-page list of your 10 most important contacts (key customers, suppliers, staff, your IT provider and your bank) with phone numbers, and keep a printed or phone-saved copy that does not depend on Microsoft. Agree on a backup way to reach staff, such as a group text.

⚡THE RIPPLE EFFECT - Three stories, and what they actually mean for your business

A B.C. College Application Portal Has Been Offline for Over a Week
EducationPlannerBC, the shared application portal used by B.C.'s public colleges and universities, has been offline for more than a week after an intrusion it detected on November 3. The organization says an unauthorized actor accessed a limited set of applicant data, including names, contact details and gender, but not Social Insurance Numbers or financial information. It follows the ransomware attack on the Burnaby School District and a run of incidents hitting Canadian education.
What it means for you: When a shared platform goes down, everyone who depends on it stops. List the outside services your business could not operate without, and ask each what their plan is for an outage lasting weeks, not hours.

Criminals Used Remote-Access Tools at IT Providers to Reach Customer Computers
Since November 11, attackers have used access to cloud-hosted ScreenConnect remote management panels to push rogue remote-access software onto computers managed by IT providers. Security firm Huntress counts at least 288 computers across 48 organizations, and says the malware also spreads to technicians' own workstations to steal logins. ConnectWise has disabled five third-party add-ons and says its own platform was not breached.
What it means for you: If an outside company manages your computers, ask them today whether they use ScreenConnect, whether they have checked for unexpected remote-access tools or admin accounts, and how fast they would tell you if they were affected.

More Than 35,000 AI-Built Apps Were Open to Anyone on the Internet
Researchers found 37,571 apps built with Grok's app builder, and 35,578 of them could be opened by anyone without signing in. About one in ten held internal business or personal data, including a hospital employee directory, an ICU patient briefing and a timesheet app that stored passwords in plain text. Sharing defaulted to anyone with the link, and AI-generated login checks failed without warning.
What it means for you: If anyone on your team has used an AI tool to build a quick app, form or dashboard, ask them to check who can open it. Convenience features often default to open.

6.4 million

Europol announced on November 12 that a nine-country operation, including Canada, seized 235 servers and 257 domains behind three major malware families, CastleRAT, Gootloader and StealC. Investigators found stolen credentials belonging to 6.4 million victims that were being sold to other criminals, including ransomware gangs. Gootloader spreads through fake legal documents on hacked WordPress sites. A takedown stops the machinery, but stolen passwords stay usable until they are changed.

✅ ONE THING TO DO THIS WEEK

Find Out What AI Apps Your Team Has Built, and Who Can Open Them

Easy AI app builders let anyone on your team create a working tool in an afternoon, such as an intake form, a staff directory or a timesheet. They also often default to sharing with anyone who has the link, and the login checks they generate can silently fail. The researchers' findings on Grok apps this week show what that looks like at scale.

Ask your team a simple question this week: has anyone built an app, form or dashboard with an AI tool? For each one, check who can open it without signing in, what information is in it, and whether it should exist at all. Switch anything that holds business or personal data to invited users only, and add a line to your acceptable-use policy about AI-built tools.

If it was built in an afternoon, it was probably shared with the world by default.

📌 QUICK BITES

  • York University disclosed that data from online forms going back to 2007 was accessed. Residence applications, surveys and similar forms were affected, and a small number of people had government ID or financial details exposed. If your website has had forms for years, find out what they still store and delete what you no longer need.

  • Loblaw is facing a proposed class action over how long it took to tell customers about its February breach. The suit says customers were not told until October. Plan your breach notification steps before you need them, including who decides and how fast.

  • AI provider Fireworks AI says its breach was wider than first reported. Attackers used stolen internal credentials to harvest customers' access keys and reach internal machines, exposing billing data and some customer datasets. If you use outside AI platforms, rotate your keys and watch your usage bills for spikes.

  • Canada and the European Union launched a new alliance on November 2 with agreements that include AI cooperation. Details of the AI arrangements were not yet available, but it signals closer alignment with EU rules. If you sell into Europe, keep a simple record of the AI tools you use.

💬 PARTING THOUGHT

Resilience is often described as a technology question. This week, it looks more like a habit of asking one question early: what happens to us if this stops?

Email stopped, and plenty of businesses discovered that their plan B was to wait. A portal went offline for over a week, and applicants had no plan B at all. These are not exotic failures. They are the ordinary ones nobody rehearsed.

You cannot prevent Microsoft's outages. You can decide in advance how you will keep serving your customers when one happens.

Resilience is what you do while you wait.

See you next Friday and stay
TryberResilient