The five decisions that matter

  • What is your share of the responsibility? → Section 2

  • Which risks should you tackle first? → Section 3

  • Separate tools or a single CNAPP? → Section 5

  • How do you incorporate security into the development pipeline? → Section 7

  • Do it yourself or outsource it? → Section 12

1. What is cloud security?

Cloud security encompasses the measures, processes, and tools that protect data, applications, and infrastructure in the cloud: from identities and access rights to networking, encryption, logging, and workload security. It differs from traditional security in that there is no longer a fixed perimeter. Resources come and go via code, and any identity with too many privileges is a point of entry.

Organizations are investing heavily in it. Gartner expects global spending on information security to grow by 12.2% to $246.2 billion in 2026. Cloud security is growing the fastest: by 28.8% to $17.1 billion.

2. The shared responsibility model

The more the service is managed, the smaller your share. Identities, configuration, and data always remain your responsibility, even with SaaS.

Low IaaS PaaS SaaS
Data Center, Hardware, and Network Provider Provider Provider
Virtualization and Operating Systems You (operating system) Provider Provider
Application and Runtime You You (application) Provider
Service Configuration You You You (configuration)
Identities and Access You You You
Data You You You

Provider secured by the cloud provider You secured by you

3. The greatest risks

 

Risk Example Measure
Misconfigurations A storage bucket or database that is publicly accessible CSPM, policies as code, guardrails in the landing zone
Excessive permissions Admin permissions for everyone, access keys that are valid for years Least privilege, CIEM, short-lived credentials, MFA
Vulnerabilities in workloads Outdated images and libraries Scanning, a fixed patching process, minimal images
Leaked secrets API keys in code or in CI logs Secrets manager, rotation, secret scanning
No logging or detection Audit logs are turned off, or no one is looking at them Centralized logging, detection, and an incident response process
Supply Chain A compromised dependency or a malicious package SBOM, signed builds, trusted registries
Shadow IT and Shadow AI GenAI tools fly under the radar of IT; 57% of employees use personal GenAI accounts for work (Gartner) Policy, an approved alternative, and monitoring

4. Zero Trust

Zero Trust assumes that no user, device, or workload can be trusted by default, not even within the organization's own network. Three principles:

  • Explicitly verify each access attempt based on identity, device, and context;
  • Grant the minimum necessary permissions for the shortest possible time;
  • Assume a breach: segment and log, so that an attacker doesn't get very far and you can detect them.

In practice: single sign-on with MFA, access to applications via an identity-aware proxy (Zero Trust Network Access) instead of a VPN, segmentation between workloads, and separate identities for workloads and AI agents. Read more in our blog about Zero Trust.

5. CSPM, CWPP, CIEM, and CNAPP

A CNAPP, such as Wiz, replaces a collection of disparate tools and highlights which combinations of vulnerabilities are truly dangerous—for example, a publicly accessible VM with a vulnerability and an identity with administrative privileges. Gartner expects that by 2029, half of all enterprise applications will run in containers, which will increase the demand for this type of integrated protection. Choose your tools based on the questions you want to answer and the people who will follow up on the alerts.

Abbreviation What it does
CSPM (Cloud Security Posture Management) Checks your cloud configuration for errors and deviations from standards such as CIS
CWPP (Cloud Workload Protection Platform) Protects workloads—VMs, containers, and serverless—from vulnerabilities to runtime
CIEM (Cloud Infrastructure Entitlement Management) Identifies identities and rights and finds rights that are too broad or unused
KSPM (Kubernetes Security Posture Management) CSPM for Kubernetes Clusters
CNAPP (Cloud Native Application Protection Platform) Combines the above into a single platform, so you can see how risks work together to form an attack path
CASB and SASE/SSE Securing SaaS usage and user access from the network

6. Kubernetes and containers

Kubernetes comes with its own risks: clusters with overly broad permissions, unprotected network traffic between pods, insecure images, and a publicly accessible API server. The most important measures are:

  • access via SSO and RBAC with least privilege;
  • network policies that block everything by default, for example, using Cilium;
  • Pod Security Standards and Policies using Kyverno or OPA Gatekeeper;
  • Scan and sign images, and allow them only from trusted registries;
  • runtime detection, for example, using Falco;
  • a standardized upgrade process for Kubernetes and the nodes.

The complete list and its relationship to NIS2 are provided in the Kubernetes guide, Section 7.

7. Security in the development pipeline

The sooner you find a bug, the less it costs. That’s why you should incorporate security into your platform’s pipeline and golden paths:

  • Scan code, dependencies, and secrets with every merge request;
  • Scan Infrastructure as Code for misconfigurations before it is deployed, for example, using Trivy or Checkov;
  • Build container images from minimal base images, scan and sign them, for example using Sigstore;
  • Create an SBOM (software bill of materials) for each release; under the Cyber Resilience Act, this will become the standard;
  • policies as code, so that the pipeline enforces the security guidelines.

You can read about how to implement this in a developer platform in the guide on platform engineering.

8. Regulations: NIS2, DORA, CRA, and GDPR

 

Rule For whom What It Takes Effective as of
Dutch Cybersecurity Act (NIS2) Essential and important entities; more than 8,000 organizations in the Netherlands Duty of care (risk management, patching, access management, vendors), reporting requirements, registration with the NCSC; the board bears ultimate responsibility 15 August 2026
DORA Financial institutions and their critical IT service providers IT risk management, testing, incident reporting, and third-party management, including an exit strategy 17 January 2025
Cyber Resilience Act Manufacturers of products with digital elements, including software Report actively exploited vulnerabilities and serious incidents: an early warning within 24 hours, a report within 72 hours, followed by a final report. Later, also security by design and updates Reporting requirement: 11 September 2026; other requirements: 11 December 2027
GDPR Anyone who processes personal data Appropriate security measures; report data breaches to the Dutch Data Protection Authority within 72 hours 25 May 2018

Under NIS2, you must report a significant incident within 24 hours via an early warning, within 72 hours via an incident report, and within one month via a final report. The government is also subject to BIO2, the successor to BIO. Have your legal counsel or compliance officer assess how these requirements apply to your organization.

9. Data sovereignty

Data stored in a European region by a U.S. provider may be subject to U.S. laws, such as the CLOUD Act. Whether this poses a risk depends on the data, your industry, and your customers. Mitigation measures range from encryption using your own keys to a European or dedicated environment for the most sensitive data. You can read about the considerations for each dataset in the guide on Cloud Migration, section 5. A Data Sovereignty Assessment can map this out in just one week. To learn how to make choices for each workload, see the page on sovereign cloud without vendor lock-in.

10. AI and Cloud Security

AI is at the top of the 2026 security agenda.

Security trends for 2026 according to Gartner

  • Monitoring agent-based AI. AI agents, no-code, and vibe coding create new attack surfaces. Give agents their own identity, least-privilege access, and logging.

  • Identities for AI agents. Identity and access management must also cover machine identities and agents.

  • AI in the SOC. AI is transforming detection and response, but it requires new skills and comes with its own costs.

  • Shadow AI. 33% of employees surveyed enter sensitive information into unapproved AI tools. Offer a secure alternative and establish guidelines.

  • Post-quantum cryptography. Gartner predicts that current asymmetric cryptography will be insecure by 2029. Start by taking stock of where you’re using which cryptographic methods.

11. Step-by-step guide

Cloud security starts with an overview: What accounts, data, and identities do you have? Next, address the issues that eliminate the most risk first, and then implement the measures on a structural basis.

In eight steps

  • Take inventory of accounts, subscriptions, clusters, data, and identities.

  • Identify risks using CSPM or a CNAPP, or have a Cloud Security Health Check performed.

  • Address the quick wins: publicly accessible resources, overly broad permissions, lack of MFA, and logging.

  • Configure identities and access according to the principle of least privilege, using short-lived credentials.

  • Ensure centralized logging, detection, and an incident response process that meets reporting deadlines.

  • Integrate security into the pipeline and the developer platform.

  • Align your measures with NIS2, ISO 27001, or DORA so that you can demonstrate compliance.

  • Test periodically and make continuous improvements.

12. Do it yourself or outsource it?

 

Do It Yourself Outsourcing (managed cloud security)
Knowledge Cloud, Kubernetes, and security expertise within your own team Provided by the partner
Monitoring In-house staff, even outside office hours As agreed with the partner
Tooling Choose, set up, and maintain it yourself Often included or set up by the partner
Direction Entirely with your own team With your own team, with the partner serving as the contractor

Many organizations retain control and policy-making in-house, while outsourcing monitoring and follow-up. To learn how to make that decision, read the guide on outsourcing IT management.

13. Frequently asked questions about Cloud Security

All frequently asked questions

The infrastructure of major providers is usually more secure than the average in-house data center. But in the cloud, a single configuration error can immediately expose data to the public. The difference lies in the design: with guardrails, least privilege, and continuous monitoring, the cloud can be secured effectively.

CSPM checks your cloud configuration for errors. A CNAPP combines CSPM with workload protection, access management, and often pipeline scanning. This shows you how separate risks combine into an attack path.

The agreement between the cloud provider and the customer regarding who is responsible for securing what. The provider is responsible for the cloud infrastructure. The customer remains responsible for identities, configuration, and data at all times, even with SaaS.

Significant incidents: an early warning within 24 hours, an incident report within 72 hours, and a final report within one month. In the Netherlands, the Dutch Cybersecurity Act has governed this since 15 August 2026.

A one-week assessment in which two consultants from SUE review the security of your cloud environment: code and pipelines, applications in production, and the infrastructure. You’ll receive a report, an opportunity and impact matrix, and a roadmap that starts with the risks that have the greatest impact. Learn more on the Cloud Security Health Check page.
Yes. A provider like SUE can set up your security tools, monitor your environment, and resolve any issues found. However, the ultimate responsibility for your data and for compliance with regulations such as the Cybersecurity Act remains with your own organization. Therefore, document who does what, how quickly an incident is reported, and how you will demonstrate during an audit that the system is working. See Managed Cloud Security.
BIO2 was adopted on 23 September 2025, and is mandatory for the national government, the provinces, and the water boards; municipalities use it as a guideline. BIO2 builds on ISO 27001 and ISO 27002:2022 and takes a risk-based approach: the fixed baseline security levels have been eliminated. In the cloud, this means: determining the risks for each system, selecting measures, and demonstrating that they are effective—including at your cloud provider. BIO2 will serve as the framework for the government’s duty of care under the Cybersecurity Act.
The Cybersecurity Act has been in effect since 15 August 2026. The duty of care requires, among other things, risk analysis, incident response, business continuity and backups, supply chain security, secure development with vulnerability management, encryption, access management with multi-factor authentication, and training. In the cloud, this involves identities and permissions, logging and detection, and the requirements you establish for your suppliers.

Where is your cloud right now?

You can see where your cloud stands right now by running a Cloud Security Health Check.

Stefan Behlen
Stefan Behlen

Let’s talk


Where is your cloud right now?

You can see where your cloud stands right now by running a Cloud Security Health Check.

* required

We will respond within one business day. By submitting this form, you confirm that you have read and understood our privacy policy.
SUE Cloud & AI Native Solutions
Privacy overview
This website uses cookies. We use cookies to ensure that our website and services function properly, to gain insight into the use of our website, and to improve our products and marketing. For more information, please read our privacy and cookie policy.