The Privileged Account Landscape
Privileged accounts are the master keys of your enterprise infrastructure. Every system, application, and cloud service relies on some form of elevated access — and each represents a potential attack vector when unmanaged. Select any account type below to explore its risk profile and Delinea's approach to securing it.
Domain Administrator Accounts
Domain administrator accounts hold the highest level of trust in a Windows environment. They have unrestricted access to all domain-joined systems, can create or modify any account, and control Group Policy across the entire forest.
Why Domain Admins are Primary Targets
- Unrestricted access to all domain-joined machines and data
- Ability to create accounts, modify Group Policy, and alter security settings
- Hash extraction via DCSync attacks grants offline credential cracking
- Pass-the-Hash and Pass-the-Ticket attacks allow lateral movement
- Kerberoasting targets service tickets tied to admin accounts
Where Organizations Go Wrong
- Using domain admin accounts for daily tasks or email
- Shared DA accounts with no individual accountability
- No separation between Tier 0 and lower-tier admin accounts
- Passwords unchanged for months or years
- Domain Admins logging into workstations (credential exposure)
Real-World Exploitation Paths
- Golden Ticket attacks via KRBTGT account compromise
- Credential dumping from LSASS memory on any domain machine
- Adversary-in-the-middle attacks on admin sessions
- Social engineering targeting IT staff with DA access
- Ransomware operators deliberately acquire DA for max impact
Local Administrator Accounts
Every Windows machine has a built-in local administrator account. When the same password is shared across thousands of endpoints, a single breach becomes an enterprise-wide lateral movement highway.
The Lateral Movement Enabler
- Same local admin password across fleet enables one-to-many compromise
- Local admin rights allow credential harvesting via LSASS
- Pass-the-Hash attacks exploit reused NTLM hashes across machines
- Attackers use local admin to install backdoors and persist
- Often neglected in patch cycles and rotation policies
Endpoint Admin Pitfalls
- Identical local admin passwords deployed via imaging tools
- Built-in Administrator account (RID 500) never renamed or disabled
- End users with full local admin rights for "convenience"
- No monitoring of local admin group membership changes
- Remote management tools (RDP, WinRM) always enabled with admin
Why This Multiplies Risk
- An enterprise with 10,000 endpoints may have 10,000 identical attack paths
- Shared hashes enable automated lateral movement at scale
- Difficult to detect without dedicated endpoint telemetry
- Compliance frameworks (PCI DSS, HIPAA, CIS) explicitly require unique credentials
- Remote work expansion has multiplied unmanaged endpoint attack surface
Service Accounts
Service accounts are non-interactive accounts used by Windows services, scheduled tasks, and background processes. Because they rarely have human oversight, they are frequently over-privileged, under-rotated, and invisible to security teams.
The Silent Privileged Identity
- Passwords set to "never expire" to avoid service disruptions
- Often granted Domain Admin rights "just in case"
- No human monitoring suspicious behavior patterns
- Credentials frequently embedded in scripts or config files
- Shadow IT services create undiscovered accounts outside IT governance
Service Account Failures
- Overprovisioning: services running as Domain Admin
- Password never rotated due to fear of breaking dependencies
- Same account used by multiple unrelated services
- Interactive logon allowed (enabling human use of service account)
- No documentation of what systems depend on the account
Why They're Hard to Find
- Created by application installers without IT knowledge
- Developers create service accounts that persist long after projects end
- Hundreds of accounts accumulated over years across AD
- Group Managed Service Accounts (gMSA) often misconfigured
- Cloud-synced accounts create hybrid blind spots
Application Accounts
Application accounts are credentials hardcoded or embedded within software to authenticate to databases, APIs, and backend systems. They represent one of the most pervasive and difficult-to-remediate privileged identity problems in modern IT.
The Hardcoded Credential Crisis
- Credentials embedded in source code, config files, or binaries
- Checked into version control repositories (GitHub, GitLab, Bitbucket)
- Accessible to any developer with repository access
- Frequently with broad database or API permissions
- Changing them requires code deployment — so they never change
Developer Security Anti-Patterns
- Connection strings with plaintext credentials in web.config or .env files
- Database accounts with DBA rights when read-only would suffice
- Generic shared application accounts with no per-service isolation
- Same credentials in dev, staging, and production environments
- API keys committed to public or semi-public repositories
How App Credentials Leak
- Code repository exposure (intentional or accidental public repo)
- Log files capturing connection strings during errors
- Memory dumps and heap snapshots exposing in-memory credentials
- Container images published with embedded credentials
- Build artifact repositories containing plaintext secrets
Emergency Break-Glass Accounts
Break-glass accounts are highly privileged emergency accounts that exist outside normal IAM workflows. They're designed for crisis scenarios — but their exceptional power and loose governance make them prime attack targets.
High Power, Low Oversight
- Designed to bypass normal MFA and conditional access policies
- Often stored on paper or in a physical safe — no digital audit trail
- Infrequently used means compromise may go undetected for months
- Typically granted the highest possible system permissions
- Cloud break-glass accounts (Global Admin) can affect entire tenants
Emergency Account Pitfalls
- Password written on physical paper with poor physical security
- Same break-glass password for years without rotation
- No alerting when break-glass accounts are used
- Unclear criteria for when break-glass access is justified
- Insufficient personnel awareness — lost credentials in emergencies
Regulatory Expectations
- NIST SP 800-53 requires monitoring and auditing of emergency accounts
- ISO 27001 demands documented emergency access procedures
- SOC 2 auditors scrutinize evidence of break-glass account governance
- PCI DSS requires immediate password change after break-glass use
- Azure AD / Entra ID best practices mandate ≥2 cloud break-glass accounts
SSH Keys
SSH keys provide cryptographic authentication to Linux, Unix, and network infrastructure. Unlike passwords, they are rarely included in rotation policies — creating massive, invisible inventories of persistent privileged access.
The Forgotten Credential Type
- Private keys stored on developer workstations without passphrase protection
- Authorized_keys files accumulate over years — "orphan" keys grant persistent access
- Keys created for temporary access (contractors, projects) never removed
- Compromised developer laptop exposes keys to all associated servers
- No built-in expiration mechanism in SSH itself
SSH Key Management Failures
- Wildcard authorized_keys entries granting access from any key
- Root login over SSH enabled (PermitRootLogin yes)
- Keys shared across multiple team members ("team key")
- Private keys checked into source code repositories
- No inventory of which keys authorize access to which servers
Where SSH Keys Live
- Linux servers (physical, virtual, and cloud)
- Network appliances (switches, firewalls, routers)
- CI/CD pipelines and build systems
- IoT and industrial control systems
- Automated backup and replication systems
API Tokens & Keys
API tokens and keys are bearer credentials that authorize machine-to-machine communication. Their power lies in their simplicity — which is also their greatest security weakness. A token in the wrong hands provides immediate, often broad access.
Bearer Tokens: Possession = Access
- No identity verification — whoever holds the token gets the access
- Often long-lived or non-expiring by default
- Broad scopes granted "just to make it work"
- Frequently leaked via logs, error messages, or repository commits
- Third-party services (SaaS, vendors) accumulate tokens over time
Token Security Failures
- Admin-scoped API keys used where read-only would suffice
- Tokens with no expiration date set during creation
- API keys committed to public GitHub repositories
- Same token shared across multiple services or teams
- No token rotation workflow — fear of breaking integrations
The Ecosystem of Tokens
- REST API keys (Salesforce, ServiceNow, Okta, Slack)
- OAuth 2.0 client secrets and refresh tokens
- Personal Access Tokens (GitHub, Azure DevOps, GitLab)
- Database connection tokens (MongoDB Atlas, Snowflake)
- Webhook secrets and HMAC signing keys
Cloud IAM Roles
Cloud IAM roles in AWS, Azure, and GCP control access to cloud-native resources at massive scale. Unlike traditional accounts, they are often assumed by services, workloads, and external identities — creating a sprawling, dynamic attack surface.
Cloud Privilege: Scale & Speed
- A single misconfigured role can expose entire S3 buckets or Azure subscriptions
- Roles assumed by Lambda functions, EC2 instances, or Kubernetes pods
- Privilege escalation via IAM PassRole, CreatePolicy, and AssumeRole chains
- Shadow roles created by developers outside cloud governance processes
- Cross-account trust relationships creating indirect access paths
Cloud IAM Anti-Patterns
- AdministratorAccess policy attached to service roles (wildcard: *:*)
- IAM users with long-lived access keys instead of role-based access
- No SCPs (Service Control Policies) limiting blast radius
- Public S3 bucket + overpermissioned role = data exfiltration path
- Root account used for routine operations without MFA
Governance Across Clouds
- AWS IAM roles vs. Azure RBAC vs. GCP IAM — three different models
- Federated identity (SAML, OIDC) creates additional trust chains
- Workload Identity Federation blurs the human/machine identity boundary
- Kubernetes RBAC adds another layer within cloud-hosted clusters
- No unified visibility without a CIEM or PAM solution