Cybersecurity
Remote and hybrid work security - a guide for organizations
Remote work security: VPN, multi-factor authentication, zero trust, BYOD, and the office cloud. Practical guidelines for companies working in a hybrid model according to NIST SP 800-207.
Łukasz DudkowskiPublished Updated 17 min read

Contents
- Why remote work needs a different approach
- Zero trust as the foundation of hybrid security
- VPN as one element, not the whole protection
- Multi-factor authentication (MFA)
- Office cloud security in the hybrid model
- SASE and SSE - security architecture for distributed work
- EDR and XDR - the workstation as its own perimeter
- Data classification as the basis of DLP
- Identity lifecycle management
- Responding to an incident on a remote device
- BYOD - when employees use their own devices
- Home network security
- Training and security culture
- Hybrid work security implementation checklist
- Where to start
- Legal basis
- Where we help
- Sources
In May 2021 the operator Colonial Pipeline shut down a fuel pipeline, which disrupted gasoline supply in parts of the south-eastern United States. The attack started with the password of an old VPN account that had no multi-factor authentication. The company’s CEO testified that it was not known how the attackers obtained the password.
Remote access launched in haste in 2020 needs a review. The data show that:
- according to the Verizon DBIR 2026, 31% of breaches start with software vulnerabilities, more often than with stolen passwords,
- according to Microsoft, multi-factor authentication can block more than 99.2% of attacks that lead to account compromise,
- according to IBM, the average global cost of a data breach in 2025 was USD 4.44 million.
This article describes how to build lasting safeguards for remote and hybrid work. It covers zero trust architecture, VPN, multi-factor authentication, the office cloud, and employees’ personal devices. Remote access to industrial systems is described in a separate guide: Remote access to ICS systems.
Why remote work needs a different approach
The perimeter model assumes that everything inside the company network is trusted. In hybrid work, employees connect from home, cafes, and hotels, using various devices, so the network boundary stops protecting. For this reason an organization needs a new model built on three changes:
- A distributed attack surface. Every home Wi-Fi network, personal laptop, and public connection is a potential entry point into the company infrastructure.
- Unapproved tools, including AI. Employees use applications that IT has not approved (shadow IT), including artificial intelligence tools (shadow AI). According to IBM 2025, one in five organizations reported a breach linked to shadow AI, and organizations with a high level of shadow AI paid on average USD 670,000 more.
- Management complexity. Different operating systems, VPNs, communication tools, and cloud services need one consistent security policy.
Zero trust as the foundation of hybrid security
Zero trust architecture, described in NIST SP 800-207, grants no implicit trust to any user or device. Every user, device, and connection is authenticated and authorized before a session starts, regardless of network location.
The three pillars of zero trust
| Pillar | Description | Example of implementation |
|---|---|---|
| Identity verification | Every login is confirmed with more than one factor. | Phishing-resistant multi-factor authentication, meaning FIDO2 keys or passkeys. |
| Least privilege | The user has access only to the resources needed for work. | Roles and access granted for the duration of a task (just-in-time). |
| Continuous validation | Device state and context are checked on every access request. | Conditional access policies and device compliance checks. |
TIP
Zero trust does not require replacing the whole infrastructure at once. Start with three steps. First, enforce MFA on all accounts with remote access. Second, deploy conditional access policies (Conditional Access in Microsoft Entra ID). Third, check device compliance before granting access to company data.
VPN as one element, not the whole protection
A VPN encrypts traffic between the employee and the company network and remains an important element of protection. A VPN alone is not enough, because the VPN appliance sits at the network edge and is reachable from the internet. According to DBIR 2026, software vulnerabilities are the most common way in for attackers.
How to secure a VPN
- Install patches for VPNs and other edge devices without delay. Maintain a fast track for critical vulnerabilities. Why: an edge device is visible from the internet, so attackers find it quickly.
- Disable split tunneling if possible. Split tunneling sends part of the traffic straight to the internet, bypassing the VPN. Why: a computer connected to the company network and the internet at the same time can become a bridge between them.
- The VPN should expose only the needed resources, not the whole network. Why: a compromised account reaches only what the VPN exposes.
- Monitor sessions. Unusual hours, locations, and data volumes are early warning signals.
- Deploy ZTNA instead of, or next to, a classic VPN (Zero Trust Network Access, access to resources after verifying the user and device). ZTNA does not expose the whole network.
WARNING
A VPN without MFA gives access to anyone who knows the password. According to the testimony of the Colonial Pipeline CEO in June 2021, the attack started with an old VPN profile that was not meant to be in use and had single-factor authentication. Enable multi-factor authentication at every remote access point and remove unused VPN profiles.
Multi-factor authentication (MFA)
MFA is the most effective single control for remote work. According to Microsoft, it can block more than 99.2% of attacks that lead to account compromise. Effectiveness depends, however, on the method of the second factor.
MFA levels
| Level | Method | Phishing resistance | Use |
|---|---|---|---|
| Basic | SMS code or voice call. | Low: an attacker can intercept the code or take over the phone number. | Minimum for non-critical systems. |
| Medium | Push notification with number matching, or a code from an app. | Medium: number matching makes it harder to push a user into approving a foreign login. | A good compromise between security and convenience. |
| High | FIDO2, passkeys, and hardware keys. | High: the login is bound to the genuine site and cannot be phished on a fake one. | Standard for privileged accounts and access to sensitive data. |
NIST SP 800-63B-4 treats the telephone channel (SMS, voice call) as a restricted authenticator. It advises that a service should consider a SIM change or number port, and it states that at level AAL2 at least one phishing-resistant option must be available.
Microsoft enforces MFA at sign-in to the Azure, Entra, and Intune admin portals (from October 2024) and to the Microsoft 365 admin center (from February 2025). From 1 October 2025 MFA also applies to creating, changing, and deleting resources through Azure CLI, PowerShell, and management APIs. The requirement applies to all accounts that perform these operations, including break-glass accounts. Organizations with a difficult environment could postpone the second phase to 1 July 2026.
Office cloud security in the hybrid model
Email, documents, communication, and task management run in the office cloud, so each of these elements needs a security configuration.
Main protection mechanisms
Conditional access (Conditional Access in Entra ID) is a central policy that decides who gets access to resources, from where, from which device, and in what context. Conditional Access requires an Entra ID P1 or P2 license. Without it, security defaults can be used. Example rules:
- MFA at sign-in from outside the company network,
- blocking access from devices that do not comply with the policy,
- requiring device compliance, reported by the device management system, before access to cloud documents,
- time-limited sessions for access from unmanaged devices.
Email protection blocks phishing, malware, and payment fraud by impersonating a supplier or a manager (BEC, Business Email Compromise). Real-time scanning of links and attachments catches harmful content after a message arrives.
Mobile device and application management (MDM and MAM) enforces encryption, blocks copying company data into private apps, and allows remote wiping of data when a device is lost.
Data loss prevention (DLP) limits accidental and deliberate sending of sensitive data by email, messaging, or cloud drive.
NOTE
Configuring office cloud security is a continuous process, because the vendor adds new features every month. The IT team should review new settings regularly and adjust policies.
SASE and SSE - security architecture for distributed work
VPN, CASB, SWG, and ZTNA protect different areas of remote work. Deployed separately they create silos, inconsistent policies, and visibility gaps. SASE (Secure Access Service Edge) and SSE (Security Service Edge) combine these functions into one cloud platform.
| SSE component | Function | What it protects |
|---|---|---|
| ZTNA (Zero Trust Network Access) | Verifies the user and device before every connection. | Replaces VPN and does not expose the whole network. |
| SWG (Secure Web Gateway) | Filters web traffic, blocks malware, and protects DNS. | Protects against harmful files downloaded by the browser and against malicious advertising. |
| CASB (Cloud Access Security Broker) | Gives visibility and control over cloud services and detects shadow IT. | Protects data in cloud services, including unapproved ones. |
| FWaaS (Firewall as a Service) | Acts as a cloud firewall for outbound traffic. | Applies network policies regardless of the employee’s location. |
An organization that is not ready for full SASE can deploy SSE alone without changing the network infrastructure.
CASB and shadow IT visibility
CASB shows which cloud applications employees use. It allows an organization to:
- detect unapproved applications and block the transfer of company data to them,
- enforce DLP policies in real time, for example block sending files labelled “Confidential” to a private cloud drive,
- detect anomalies in user behaviour, for example mass downloading of documents at night,
- enforce encryption of data at rest and in transit.
EDR and XDR - the workstation as its own perimeter
In the hybrid model an employee’s laptop works outside the protected company network and is its own protection perimeter. A traditional signature-based antivirus is not enough, because the organization needs real-time visibility into what happens on the device.
EDR versus XDR
| Layer | EDR | XDR |
|---|---|---|
| Scope | Endpoints: laptops and servers. | Endpoints, email, cloud, identity, and network. |
| Detection | Suspicious processes, file operations, and registry changes. | Correlation of events from many sources. |
| Benefit for remote work | Remote isolation of an infected device and incident analysis. | Detection of an attack spread over time: from a phishing message, through credential theft, to lateral movement in the network. |
WARNING
Devices of remote employees need EDR. Without it the security team cannot see what happens on devices outside the company network. The scope of visibility depends on the license, so before buying, check in the license description which EDR functions the current plan contains.
Monitoring and detection for remote work
A SIEM or XDR system should detect the following scenarios:
- Impossible travel: a login from Warsaw and 10 minutes later from Brazil.
- Volume anomalies: mass downloading of files from a cloud drive.
- New devices: a login from an unknown device to a privileged account.
- Token replay: use of a stolen session token after MFA was passed, an AiTM attack (the attacker sits in the middle of the login).
- PowerShell and WMI anomalies: suspicious activity on a device outside working hours.
Data classification as the basis of DLP
DLP needs data classification, because without it the system does not know which data to protect. Sensitivity labels mark data by confidentiality level and automatically enforce the matching safeguards.
Label taxonomy
| Label | Protection | Examples of data |
|---|---|---|
| Public | No restrictions. | Marketing materials, public price lists. |
| Internal | Sharing outside the organization is blocked. | Operating procedures, internal communication. |
| Confidential | Encryption, DLP, and a block on copying to unmanaged devices. | Customer data, contracts, offers, salaries. |
| Highly confidential | Encryption, watermark, access tracking, and a ban on printing. | Intellectual property, cryptographic keys, merger and acquisition data. |
TIP
Limit the number of labels: according to Microsoft, effectiveness drops noticeably when a user has more than five main labels or more than five sublabels. Start with one set of high-risk data, run automatic labelling in simulation mode, measure the reduction in exposure, and then scale.
Sensitivity labels, DLP policies, and conditional access form a consistent chain of data protection from classification to enforcement. An example rule: files labelled “Confidential” are available only from managed devices.
Identity lifecycle management
Conditional access decides about access at a given moment, and the full picture requires managing identity from hiring to the employee’s departure.
| Stage | Control | Mechanism |
|---|---|---|
| Joiner (new employee) | Roles and access are assigned automatically based on the position. | Automation of HR workflows in the identity directory. |
| Mover (change of position) | Permissions are updated when the position or department changes. | Access reviews and privileged access management. |
| Leaver (departure) | The account is deactivated, tokens are revoked, and data on the device is wiped remotely. | A departure procedure and remote device wipe. |
Privileged Identity Management (PIM) grants administrative permissions for the duration of a task, with justification and approval. This eliminates standing privileges, meaning always-active administrative accounts waiting to be taken over.
Access reviews mean that managers confirm or withdraw their subordinates’ access at a frequency set in the organization’s policy. Access reviews in Entra require Entra ID Governance or Entra Suite licenses. The reviewer’s lack of response can be configured as no change, removal, approval of access, or taking the system’s recommendation. Removing access on no response together with automatic application of results can cut off all users, so choose the variant with care.
Responding to an incident on a remote device
With a compromised laptop of a remote employee, physical access to the device is impossible, so the team uses remote tools. The order of actions:
- Network isolation. The EDR tool cuts the device off from the network and keeps the connection to the management console.
- Token revocation. Revoke all of the user’s sessions and tokens in the identity directory.
- Account lockout. Block the account and force an MFA reset.
- Remote analysis. A remote session in the EDR tool allows running analysis scripts, collecting artifacts, and reviewing processes in memory.
- Data wipe. For a personal device (BYOD), selectively remove company data, and for a company device wipe the entire device.
TIP
Rehearse the “compromised remote laptop” scenario in a tabletop exercise. The exercise establishes who decides about isolation, in what time, how the team communicates with an employee whose email is blocked, and whether the company has a spare device.
BYOD - when employees use their own devices
Bring Your Own Device (BYOD) lowers hardware costs and raises risk. The organization must establish how to protect company data on a device it does not fully control.
BYOD approaches
| Approach | How it works | Advantages | Limitations |
|---|---|---|---|
| MAM (Mobile Application Management) | Management at the application level: company data is encrypted in a container. | The user’s privacy is preserved and deployment is quick. | Control over the device is limited. |
| MDM (Mobile Device Management) | Full device management: encryption, password policies, and remote wipe. | Control is greatest, which makes meeting regulatory requirements easier. | Employees may be reluctant to enrol personal devices. |
| Virtual desktop (VDI) | The employee connects to a virtual environment and data never leaves the cloud. | Data is not stored locally. | Needs a stable link and brings higher infrastructure costs. |
TIP
For organizations starting a BYOD programme, the simplest starting point is MAM with application protection policies. It protects company data in the email, messaging, and cloud drive apps without interfering with the private part of the device.
Home network security
A home Wi-Fi network is sometimes the weakest link of protection. A router with a default password, outdated software, and modern encryption switched off makes attacks on work devices easier.
Home network security checklist
- The default router administrator password is changed.
- WPA3 encryption is on, with WPA2-AES as the minimum.
- Work devices run on a separate network (VLAN or SSID).
- The router software is updated to the latest version.
- The WPS (WiFi Protected Setup) function is off, because it is vulnerable to brute-force attacks.
- Remote management of the router from the internet is off.
- The Wi-Fi password is long and unique.
- A VPN at the router level covers all work traffic, if the organization’s policy requires it.
Training and security culture
Technical safeguards do not work if employees do not understand the threats. Phishing remains a frequent attack vector against remote employees. A security awareness programme consists of five elements:
- Onboarding training covers rules of remote work, recognising phishing, and handling incidents.
- Phishing simulations serve learning, not punishment, and their results show areas that need support.
- Clear reporting procedures tell an employee whom to report a suspicious email or incident to.
- Short messages remind of the rules at regular intervals set in the organization’s policy.
- Password management means using a password manager and long, unique passwords for every account.
Hybrid work security implementation checklist
The list helps organizations that plan or verify safeguards for remote and hybrid work.
Infrastructure and access
- VPN or ZTNA requires MFA at all remote access points.
- Conditional access policies are configured and tested.
- FIDO2 or passkeys are deployed for privileged accounts.
- VPN sessions are monitored for hours, locations, and volume.
- A procedure exists for fast patching of VPNs and edge devices.
Devices and data
- A BYOD policy is defined and communicated to employees.
- MAM or MDM is deployed on devices with access to company data.
- Disks of all devices are encrypted (for example BitLocker or FileVault).
- DLP policies are configured for email, messaging, and the cloud drive.
- Company data has automatic backups in the cloud.
People and processes
- Security training takes place regularly, at intervals set in the policy.
- Phishing simulations take place regularly, at intervals set in the policy.
- All employees know the incident reporting procedure.
- A clean desk and screen locking policy is in force.
- Reviews of access permissions take place regularly, at intervals set in the policy.
Where to start
Hybrid work security is built in stages. The order below starts with the change that has the greatest impact:
- Immediately: MFA on all accounts with remote access. Why: a single stolen password is then not enough to enter the network.
- Within a month: conditional access and device compliance checks.
- Within a quarter: BYOD with MAM, DLP, and a training programme.
- Continuously: monitoring, patches, permission reviews, and policy updates.
Remote access to industrial environments (ICS/OT) follows different rules, described in the guide Remote access to ICS systems. Compromised remote access is also one of the routes of ransomware attacks, as the ransomware guide describes.
Legal basis
| Step / topic | Where it is |
|---|---|
| Multi-factor or continuous authentication as a risk-management measure | NIS2 Directive, Article 21(2)(j) |
| Basic cyber hygiene and training | NIS2 Directive, Article 21(2)(g) |
| Transposition of NIS2 into national law | National law of the Member State, deadline 17 October 2024 |
Where we help
Penetration tests check the configuration of VPN, MFA, and conditional access policies and employees’ resistance to phishing. The NIS2 audit assesses whether remote work safeguards meet the requirements for essential and important entities. The SEQRED team also helps design a zero trust architecture matched to the organization.
Sources
- NIST SP 800-207 - Zero Trust Architecture - NIST (accessed: )
- Verizon 2026 Data Breach Investigations Report - Verizon (accessed: )
- IBM Cost of a Data Breach Report 2025 (press release) - IBM (accessed: )
- NIST SP 800-63B-4: Digital Identity Guidelines, Authentication and Authenticator Management - NIST (accessed: )
- Plan for mandatory Microsoft Entra multifactor authentication (MFA) - Microsoft Learn (accessed: )
- Create an access review of groups and applications - Microsoft Learn (accessed: )
- Learn about sensitivity labels - Microsoft Learn (accessed: )
- NIS 2 Directive, Article 21: Cybersecurity risk-management measures - nis-2-directive.com (accessed: )
- Colonial Pipeline CEO: Ransomware Attack Started via Pilfered 'Legacy' VPN Account - Dark Reading (accessed: )
- NIS2 Directive - European Commission (accessed: )
Need help in this area?
Our experts will help you assess the risk and plan next steps.