Welcome to ISACA Denmark Members
Welcome to our February 2026 update! This month, Denmark has experienced a shift from conventional cybercrime to direct geopolitical cyber pressure. OpDenmark—a coordinated Russian-led campaign—has made Denmark the target of extensive DDoS attacks with a clear political purpose: to pressure the Danish government to stop military support to Ukraine. Simultaneously, we're seeing the beginning of a new technological era where autonomous AI agents are becoming both weapons and defenders in cyber warfare.
For Danish cybersecurity professionals, February 2026 marks a turning point. We're no longer in a world of random hacktivism, but facing structured, repeatable attack playbooks targeting our municipalities, welfare portals, and critical infrastructure. This article examines the OpDenmark campaign, introduces the concept of "Know Your Agents" (KYA) as the next phase of Identity and Access Management, and explains why the next wave of attacks against Denmark will be run by AI agents, not humans.
OpDenmark: Cheap Coercion Against a Digital State
February 2026 was dominated by an aggressive cyber campaign launched by a pro-Russian hacker alliance called "Russian Legion" and affiliated groups. OpDenmark, as the campaign was named, represents a new form of hybrid warfare where cyberattacks are explicitly used as diplomatic pressure.
Campaign Structure: OpDenmark launched large-scale DDoS (Distributed Denial of Service) attacks against Danish companies and public organizations, with the energy sector repeatedly highlighted as a primary target. The attackers first threatened the campaign and then systematically executed the attacks, demonstrating coordination and planning far beyond opportunistic hacktivism.
The most alarming aspect was the campaign's explicit political demands: attackers demanded that Denmark halt a 1.5 billion DKK military aid package to Ukraine. This marks a direct link between cyberattacks and foreign policy decisions—cyber warfare as a diplomatic coercion tool.
Greenlandic Dimension: The OpDenmark campaign also hit Greenlandic public websites with DDoS attacks. The Danish Agency for Social Security confirmed ongoing attacks and active coordination between Danish and Greenlandic authorities to handle the threat. This underscores Denmark's and Greenland's vulnerability as small, highly digitalized states where even relatively simple DDoS attacks can disrupt critical public services.
Defense Response: In response to the escalating cyber threat, Denmark's Defense Intelligence Service (FE) publicly intensified recruitment of cybersecurity specialists for offensive cyber operations. This signals a more assertive national cyber posture, where Denmark is moving from primarily defensive to more proactive cyber defense, including offensive capabilities.
Implications for Q1 2026 and Beyond: OpDenmark is not a one-off incident but the beginning of a new normal. As long as Denmark continues its support to Ukraine—which all political signals suggest—we can expect:
- Sustained DDoS campaigns against public authorities, municipalities, energy supply, and media
- Escalation from nuisance to serious disruptions of critical services
- More open use of offensive cyber tools by Danish defense and intelligence actors
- Closer NATO-coordinated cyber-military integration to counter state threats
For Danish organizations, especially those in critical infrastructure, this means:
-
DDoS preparedness is no longer optional: Organizations must have tested mitigation plans, redundant systems, and clear communication strategies for prolonged attacks.
-
NIS2 implementation gains acute relevance: The EU's NIS2 directive, which Denmark is implementing, requires precisely the type of resilience and incident handling that OpDenmark demonstrates the need for.
-
Public-private collaboration is critical: Effective response requires coordination between authorities, sector actors, and the cybersecurity community to share threat intelligence and coordinate defense.
From Random Hacktivism to Structured Playbooks: OpDenmark marks a shift from opportunistic attacks to structured, repeatable campaigns. Attackers use known playbooks—threat announcements, coordinated DDoS waves, sector-specific targeting—that can be activated in response to political decisions. This makes the cyber threat predictable in timing (tied to diplomatic events) but not necessarily in scope or effect.
For Danish cybersecurity leaders, this means risk assessment must now include geopolitical analysis: Which political decisions can trigger cyberattacks? How does Denmark's international position affect our threat landscape?
From OpDenmark to Agentic Attacks: AI as the Next Wave
While OpDenmark represents today's threats, we're already seeing the contours of tomorrow's attacks: autonomous AI agents that can plan, execute, and adapt cyberattacks without continuous human control.
The Agentic AI Revolution: 2026 is the year when agentic AI is moving from experiments to production on both attacker and defender sides. An agentic AI is not just a tool but an autonomous agent that can:
- Plan complex attack campaigns based on objectives and available resources
- Execute actions across cloud, identity systems, and infrastructure
- Adapt to defenses in real-time, learn from failed attempts, and shift tactics
- Orchestrate entire kill chains from reconnaissance to exfiltration
Global cybersecurity leaders expect AI agents will soon:
- Manage cloud resources autonomously
- Generate real-time phishing tailored to individual victims
- Orchestrate complete attack chains without human intervention
- Multiply attack surfaces by identifying and exploiting vulnerabilities faster than humans can patch them
Danish Context: For Denmark, this means a new generation of threats against:
- E-ID and NemID/MitID systems: Automated credential stuffing and sophisticated phishing driven by AI agents
- Municipal services: Autonomous attacks that identify and exploit vulnerabilities in citizen portals and welfare systems
- Energy supply: AI-orchestrated attacks against SCADA systems and critical infrastructure
- Financial systems: Real-time fraud agents that adapt to fraud detection systems
Simultaneously, governments and SOC providers are adopting AI agents for detection, containment, and triage—"agentic SOCs" that can respond to threats at machine speed.
The Strategic Dilemma: Denmark faces an asymmetric challenge: attackers need only one successful AI agent to compromise critical infrastructure, while defenders must protect thousands of potential attack points. This creates a race to develop and deploy defensive AI agents faster than attackers can innovate offensive capabilities.
Know Your Agents (KYA): The Missing Pillar in IAM
As AI agents become ubiquitous—both as productivity tools (Microsoft Copilot, custom LLM agents) and as infrastructure automation—a critical security problem emerges: AI agents are the most dangerous insider threat if not properly governed.
Agents are:
- Always active: They operate 24/7 without human supervision
- Highly privileged: They often have broad access to automate effectively
- Implicitly trusted: Organizations assume their own agents are safe
This creates new vulnerabilities: rogue agents, poisoned models, and cascading failures when one orchestrator agent fails or is compromised.
Introducing KYA: Know Your Agents
"Know Your Agents" is 2026's equivalent of KYC (Know Your Customer) and KYB (Know Your Business) for AI. It represents a new governance paradigm where AI agents are treated as a distinct class of privileged identities in Identity and Access Management (IAM) systems.
The Six Pillars of KYA
1. Agent Identity Inventory Every AI/automation agent—from Copilot-style tools to custom LLM agents, RPA bots, and infrastructure orchestrators—must have:
- Unique identity registered in the IAM system
- Documented owner (individual or team responsible for the agent)
- Business purpose and use case
- Technical architecture and dependencies
Danish practice: "If you don't know which agents are running in your organization, you've already lost control. Inventory is the foundation for all other governance."
- Narrowly scoped roles (per application, per environment, per data domain)
- Short-lived credentials, like modern workload identities
- Just-in-time access provisioning for critical operations
- Automatic rotation of secrets and API keys
Danish practice: "Treat agents as zero-trust workloads: least privilege, explicit verification, assume compromise."
3. Segregation of Duties for Autonomous Workflows High-risk actions initiated by agents require controls:
- Human approval for critical operations (creating privileged accounts, changing MFA rules, deployment to production)
- Separate control agents that verify other agents' actions
- Dual-control workflows for sensitive data extraction or configuration changes
Danish practice: "If an agent can change security policies or give itself more access, you've built a backdoor."
4. Continuous Behavior Monitoring ("Agent UEBA") Treat agents as super-privileged users in User and Entity Behavior Analytics:
- Baselines for what agents should access
- Anomaly detection when they start accessing new tenants, countries, or datasets
- Pattern recognition that identifies compromised or rogue agents
- Real-time alerting on deviant behavior
Danish practice: "An agent that suddenly accesses HR data at 3 AM when it normally only runs business hours reports should trigger an alarm."
5. Data Provenance and Model Integrity Checks 2026 attackers focus more on corrupting training data and models than stealing data:
- Signed models and policies to prevent tampering
- Change control around prompts, tools, and connectors
- Data provenance tracking to validate training data sources
- Model testing before deployment to detect backdoors or bias
Danish practice: "Data poisoning is the new backdoor. Verify what your agents are learning from."
6. Lifecycle and Off-boarding for Agents Formal processes for agent lifecycle:
- Creation governance: Approval workflow before new agents are deployed
- Periodic review: Regular validation that agents are still necessary
- Decommissioning: Automatic revocation of secrets and access when projects end
- Audit trails: Complete logging of agent creation, modification, and deletion
Danish practice: "Experiment agents from proof-of-concepts shouldn't run in production for years with unchanged credentials."
Why KYA Is Critical for Danish Organizations
Danish organizations have particular challenges that make KYA urgent:
Digital welfare state vulnerability: Denmark's high digitalization means compromised agents can affect millions of citizens through welfare systems, health portals, and municipal services.
NIS2 and upcoming regulation: The EU's NIS2 directive and upcoming AI regulation will likely require documentation and control of autonomous systems. KYA principles position organizations for compliance.
Limited cyber workforce: Denmark's cybersecurity skills gap (which we discussed in the January article) means autonomous agents become necessary to scale security operations—but poorly governed agents worsen the problem.
State-sponsored threat: As OpDenmark demonstrates, Denmark is a target for sophisticated state actors. These actors will exploit poorly governed AI agents as attack vectors.
Implementing KYA: Practical First Steps
For Danish cybersecurity leaders wanting to implement KYA:
Week 1-2: Discovery and Inventory
- Identify all existing AI/automation agents in the organization
- Document ownership, purpose, and technical architecture
- Map existing access and privileges
Week 3-4: Risk Assessment
- Classify agents by risk (data access, privilege level, criticality)
- Identify agents with excessive access or unclear business value
- Prioritize high-risk agents for rapid remediation
Month 2: Quick Wins
- Implement credential rotation for existing agents
- Remove or downgrade agents without clear business value
- Establish logging and monitoring for high-risk agents
Month 3-6: Governance Framework
- Integrate agent identities into IAM systems
- Implement approval workflows for new agents
- Establish UEBA or anomaly detection for agent behavior
- Define policies for agent lifecycle
Ongoing operation: Treat KYA as an ongoing governance process, not a one-time project.
The Three Questions KYA Must Answer
The core of Know Your Agents can be distilled into three fundamental questions that every organization must be able to answer in real-time:
- Which agents do we have? (Inventory and classification)
- What are they allowed to do? (Privileges, access, and policies)
- What are they doing right now? (Monitoring, anomaly detection, and audit)
If you cannot answer these three questions for your AI agents, you have a blind spot in your security architecture—a blind spot that both criminals and state actors will exploit.
The Future Threat Landscape: Agentic Attacks Against Denmark
Let's look concretely at what future AI-driven attacks against Denmark could look like:
Scenario 1: Autonomous Credential Stuffing Against MitID An attacker deploys thousands of AI agents that:
- Collect compromised credentials from data breaches
- Identify Danish users through OSINT
- Test credentials against MitID with intelligent rate-limiting bypass
- Adapt timing and methods based on detection responses
- Automatically escalate successful logins to privilege escalation
Scenario 2: AI-Orchestrated Phishing Against Municipalities Agents that:
- Scrape municipal websites and social media for staff information
- Generate personalized phishing emails based on current local events
- Adapt social engineering based on victim response
- Coordinate multi-stage attacks across municipal systems
- Learn from failed attempts and continuously improve
Scenario 3: Autonomous Energy Grid Reconnaissance Agents deployed by state actors that:
- Systematically map Danish energy supply systems
- Identify SCADA vulnerabilities through automated scanning
- Test access controls without triggering alarms
- Build detailed attack maps for future use
- Remain latent until activation during geopolitical crisis
These scenarios are not science fiction—the technology already exists. The question is not if, but when we see such attacks against Denmark.
Defense AI Agents: The Other Side of the Coin
It's important to acknowledge that AI agents are not only a threat—they're also our most promising defense. Danish organizations and authorities are already developing defensive AI capabilities:
Agentic SOCs: Security Operations Centers that use AI agents to:
- Analyze millions of logs in real-time
- Identify anomalies faster than human analysts
- Automatically respond to known threat patterns
- Coordinate response across systems and organizations
Adaptive Threat Hunting: Agents that:
- Proactively search for indicators of compromise
- Predict likely attack vectors based on threat intelligence
- Continuously test defensive controls
- Simulate attacks to identify weaknesses
Automated Incident Response: Systems that can:
- Isolate compromised systems in seconds
- Revoke credentials and access automatically
- Collect forensic evidence while attacks are ongoing
- Coordinate communication to stakeholders
The Defense Intelligence Service's recruitment of offensive cyber specialists suggests Denmark is also developing offensive AI capabilities—agents that can identify, exploit, and report on adversary systems.
Conclusion: A Turning Point for Danish Cybersecurity
February 2026 marks two critical turning points for Danish cybersecurity:
Geopolitical: OpDenmark demonstrates that Denmark can no longer treat cyber threats as primarily criminal or opportunistic. We are targets for structured, politically motivated cyber warfare that will continue as long as Denmark pursues its current foreign policy.
Technological: The emergence of agentic AI fundamentally changes both attacks and defense. AI agents will within months be able to orchestrate attacks that today require teams of specialized hackers—and they'll be able to do it continuously, without human limitations.
For Danish cybersecurity professionals, this requires a dual response:
-
Short term: Strengthen DDoS preparedness, accelerate NIS2 implementation, and build robust public-private partnerships for threat intelligence sharing.
-
Medium term: Implement Know Your Agents as a governance framework, integrate AI agents into IAM systems, and build capabilities to detect and respond to agentic attacks.
ISACA Denmark continues to be a critical platform for knowledge sharing, best practice development, and community building as we navigate these challenges. Our collective expertise and collaboration are crucial for building Denmark's resilience in this new era of geopolitical cyber warfare and AI-driven threats.