Back to the list of entries

Outlook Was Down for a Day. Where Did Your Data Go?

On Thursday, 12 November, shortly before 1 p.m. Eastern time, Microsoft 365 users started losing access to their mail. Within the hour Teams, SharePoint, OneDrive and Copilot were affected as well, and Downdetector had logged around 130,000 reports about Outlook alone. Recovery was slow. On Friday morning Microsoft was still reporting partial restoration and warning that customers outside the Americas could be affected too. The company attributed the incident to a number of servers becoming unexpectedly unavailable and said it had found no sign of malicious activity.

The press has mostly counted the cost in lost working hours. We would like to look at a less visible consequence, one that tends to surface weeks later during an audit or a compliance review: the data employees moved while the corporate tools were unavailable.

Any cloud provider can have a bad day, and Microsoft is far from the only one to have had one this year. The concern applies to every organisation whose daily work runs through a single productivity suite.

How employees work around an outage

When email stops working, the work it carries doesn't stop with it. A signed contract still has to reach the client. A pricing proposal for a tender still has to be submitted by the end of the day. A project manager has a partner call in an hour and the design files are sitting in a SharePoint library nobody can open, although someone has a local copy.

Conscientious employees find a way. The contract goes out from a personal Gmail account. The spreadsheet lands in a Telegram chat with a colleague, or in a WhatsApp group with the client. A presentation is uploaded to a free file-sharing service and the link is sent by text message. Someone copies a project folder to a USB stick to finish the work at home, and someone else, with Copilot unavailable, pastes the draft of a confidential report into a public AI chatbot to get the summary written.

Nobody involved is trying to cause harm. They are trying to meet a deadline. But within a few hours a company of a thousand people can end up with sensitive documents scattered across personal mailboxes, private messenger histories, public links and removable drives, none of which the organisation controls or even knows about.

Why the risk peaks while the service is down

Many organisations rely on protection that is part of the cloud suite itself: mail-flow rules, sensitivity labels, sharing restrictions in the collaboration platform. These work well as long as data travels through the platform. During an outage it travels somewhere else, so those controls simply never see it.

Time pressure makes things worse. People in a hurry send the whole folder rather than the single page that was needed, and they pick the fastest channel available rather than the safest one.

Most importantly, the improvised copies stay where they ended up. When the service comes back, everyone returns to Outlook and Teams, but the contract is still in a personal mailbox, the customer list is still in a chat on someone's phone, and the file-sharing link is still active. For organisations subject to GDPR, India's DPDP Act, South Africa's POPIA, the UAE's PDPL or Türkiye's KVKK, a few hours of improvisation can turn into a data incident that has to be explained to a regulator.

What DLP can do when the cloud is unavailable

If data protection depends on the cloud service being reachable, it fails at exactly the moment the service does. Controls that run on the employee's computer don't have that problem.

An on-premises DLP system with an agent installed on each workstation checks data at the point where the employee tries to send it or copy it. The agent doesn't need Outlook, Teams or the cloud tenant in order to work. It analyses the file or text, applies the organisation's policy and allows, blocks or records the action locally.

In practice this means a confidential contract can be stopped on its way to a personal webmail account while the employee's ordinary browsing carries on untouched. A spreadsheet with customer personal data can be prevented from leaving through Telegram or WhatsApp. Copying sensitive files to an unapproved USB drive can be blocked while encrypted corporate drives remain available. The same content-aware rules apply to file-sharing sites and to the web versions of public AI services. Every attempt is logged, so once the outage is over the security team can see what was tried, by whom, and whether anything actually left.

The decision depends on what the data is, not only on which channel is being used. Employees can still use the tools they need, and the material that has to stay inside the company does.

How Zecurion approaches the problem

We built Zecurion Next Generation DLP on the assumption that protection should stay with the data and the people working with it, whatever happens to the platforms around them.

Zecurion controls more than 100 data transfer channels and over 250 internet services, including personal webmail, public messengers such as WhatsApp and Telegram, cloud storage and file-sharing sites, so the channels people improvise with fall under the same policies as corporate email. Zecurion Device Control manages USB drives and other removable media, permitting approved corporate devices and blocking the rest.

Agents carry separate policy sets for online and offline operation. A workstation that loses its connection to the management server keeps enforcing the rules defined for that situation rather than defaulting to allowing everything. To recognise sensitive content, Zecurion combines twelve detection techniques, among them digital fingerprints, regular expressions, machine learning, AI image recognition and OCR. As a result a confidential report is still identified after it has been renamed, converted to PDF or photographed from the screen.

The system can be deployed in 13 different ways, including fully on-premises, so organisations that want their data protection to be independent of any external cloud can keep all of it within their own infrastructure. User behaviour analytics and risk-based assessment flag unusual activity, such as a sudden burst of uploads or files going to addresses a user has never written to before, so the security team can focus on the events that matter on what is already a difficult day.

Before the next outage

There will be another outage, whether with this provider or another one. Organisations that come through it without a data incident are usually the ones that decided in advance what employees should do.

Start by agreeing on a sanctioned fallback channel and telling staff about it. Blocking every workaround without offering an alternative just pushes people towards less visible ones. Then check which of your data protection controls depend on the cloud service being reachable, and make sure each one has an equivalent on the endpoint. Define what should be blocked, allowed or only logged when workstations can't reach their usual services, and test those settings before you need them. Limit removable media to approved, encrypted devices. During the incident, keep an eye on transfers to personal email, messengers and file-sharing services, because that is when unusual activity matters most. Once services are restored, use the DLP logs to find out what moved and where, revoke or recover what you can, and decide whether a notification is required.

This week's outage will be remembered as a lost day of work. It is also worth treating as a test of where your data protection actually runs, before an incident makes the result harder to ignore.

To see how Zecurion Next Generation DLP keeps working when cloud services don't, contact the Zecurion team or your local partner for a demo.

Tags by post

cybersecurity tips

Subscribe to our blog updates

You will receive only really useful emails and will always be able to unsubscribe from this mailing if, suddenly, your interests change

Recommended resources