Steps to Revoke and Rotate Compromised TLS Certificates

A compromised TLS certificate can allow an attacker to impersonate a service, intercept traffic, or establish trust with systems that rely on certificate validation. The risk becomes more serious when the associated private key has been exposed, when certificate use is poorly documented, or when the same credential protects several applications.

Revocation and replacement should therefore be treated as an incident response activity, not a routine renewal task. Australian organisations may need to coordinate teams across Sydney, Melbourne, Brisbane, Perth, or offshore operations while maintaining customer access, regulatory reporting, and critical services. A controlled process limits the window of abuse and produces evidence that the response was properly managed.

Confirm The Certificate Compromise

Begin by establishing what has actually been exposed. A certificate is public information in most TLS deployments; the private key is the sensitive asset. Evidence of a leaked certificate file, unauthorised key access, suspicious authentication, a compromised build server, or an attacker-controlled endpoint should be treated as a potential private-key compromise until investigated.

Identify the certificate’s subject, SAN entries, issuer, serial number, thumbprint, validity period, key algorithm, and deployment locations. Search certificate stores, secrets managers, load balancers, web servers, API gateways, Kubernetes ingress objects, service meshes, reverse proxies, mobile applications, and partner integrations. Certificate Transparency logs can reveal publicly trusted certificates that were issued unexpectedly, while cloud audit logs may show who accessed or exported a private key.

Preserve evidence before making changes where practical. Record relevant timestamps, log sources, certificate files, key fingerprints, and the reason for the emergency action. If the exposed key belongs to a certificate authority, signing service, code-signing system, or high-value internal PKI, escalate immediately because the response may involve a wider trust-chain review.

The incident team should also decide whether the certificate is being actively abused. Indicators can include unexpected TLS handshakes, fraudulent domains using similar names, unusual traffic through a cloud load balancer, or reports from customers and threat intelligence providers. Active exploitation may require containment measures before the replacement certificate is ready.

Revoke The Old Certificate

Contact the issuing certificate authority through its emergency revocation process. Provide the certificate serial number and select the correct reason, such as key compromise or unauthorised use. Revocation is usually published through a Certificate Revocation List, Online Certificate Status Protocol, or both. The CA may require domain validation or account verification before accepting the request.

Revocation does not instantly terminate every existing connection. Browsers, operating systems, libraries, appliances, and embedded devices handle CRL and OCSP information differently, and some environments use soft-fail behaviour when status checking is unavailable. Internal services may not check public revocation data at all. For that reason, revocation must be paired with removal of the old certificate and private key from every reachable system.

Disable or restrict the compromised credential in the relevant secrets manager, HSM, certificate management platform, or cloud account. Replace access policies that allowed export, rotate associated API tokens, and review service identities that could have read the key. If a certificate was stored in source control, container layers, build artefacts, ticket attachments, or shared drives, remove the exposed material and invalidate any copies rather than assuming deletion is sufficient.

A broader identity review is often appropriate because certificate theft can be part of a larger compromise. CARM Security’s IAM policy review provides useful context for examining excessive permissions, cloud access paths, and post-breach control gaps that may have enabled the exposure.

Issue And Deploy The Replacement

Generate a new private key in a trusted location, preferably an HSM or managed key service with export disabled. Do not create the replacement by copying or reusing the compromised key pair. Select the strongest algorithm supported by the application estate, commonly an appropriate elliptic-curve key or RSA with a suitable key length, while checking compatibility with older clients, appliances, and partner systems.

Create a new certificate signing request with carefully reviewed subject alternative names. Avoid adding names that are not required, especially wildcard entries that would expand the impact of a future compromise. Confirm that the certificate chain, intermediate certificates, key usage, extended key usage, and validity period match the service’s role. For internal mutual TLS, verify both server and client authentication requirements.

Deploy the replacement through a controlled sequence. Many teams first install it alongside the old certificate, configure the listener to use the new credential, and test from internal and external networks. Validate the certificate chain, hostname matching, protocol negotiation, cipher configuration, OCSP stapling, session resumption, and client authentication. Test the actual business functions that depend on TLS, including payment flows, APIs, customer portals, VPNs, and automated integrations.

A blue-green or dual-certificate approach can reduce disruption when a platform supports it. For systems that cannot hold two certificates, arrange a maintenance window and define a rollback plan using only a verified, uncompromised credential. Avoid rolling back to the exposed key simply because the new deployment fails. If the private key was compromised, rollback to that key restores the original risk.

Response activity Primary purpose Evidence to retain Common failure
Confirm compromise Establish scope and urgency Logs, fingerprints, incident timeline Treating a public certificate as the secret
Request revocation Tell the issuer to invalidate trust CA case number, serial number, reason Assuming revocation reaches every client immediately
Generate new keys Break dependence on the exposed credential Key-generation record, CSR, approval Reusing the compromised private key
Deploy replacement Restore protected service access Change record, test results, deployment logs Updating one endpoint while missing others
Validate and monitor Detect residual exposure or outage Scan results, alerts, stakeholder sign-off Closing the incident before checking integrations

Verify Every Dependency

After deployment, scan the environment for the old certificate and its fingerprint. Check internet-facing hosts, DNS names, CDN configurations, WAF policies, ingress controllers, mail gateways, application servers, development environments, disaster recovery sites, and backup images. A certificate can remain active on a forgotten staging endpoint even after the main production service has been updated.

Inspect certificate inventories and use automated discovery where possible. Useful sources include cloud certificate managers, configuration management databases, vulnerability scanners, endpoint inventories, container registries, and Certificate Transparency monitoring. Compare findings against the incident scope, then confirm whether each instance is retired, replaced, or intentionally retained for a documented reason.

Pay particular attention to machine-to-machine connections. An API consumer may pin the old certificate or trust a private CA chain. A mobile application may include a public-key pin. A manufacturing device or healthcare appliance may have a long update cycle. These dependencies require coordinated changes, not an assumption that browser behaviour represents the whole environment.

Recommendations for a safer recovery process include:

Australian organisations should map these controls to their operational and regulatory context. APRA-regulated entities may need evidence aligned with CPS 234 obligations for information security capability and incident response. Businesses outside that sector can still use the ACSC Essential Eight as a practical baseline for patching, privileged access, application control, and logging. A managed detection and response provider can help correlate certificate events with endpoint, identity, and network activity when internal teams are handling several incidents at once.

Monitor The Environment After Rotation

Certificate replacement is complete only when the organisation has demonstrated that the old trust path is no longer useful. Run external TLS scans from several locations and test internal services from representative clients. Confirm that the new certificate is presented consistently through CDNs, regional load balancers, IPv4 and IPv6 endpoints, and alternate DNS records. A service operating from an AWS or Azure region in Sydney may still expose a different credential through a disaster recovery endpoint in Melbourne or Singapore.

Watch for failed connections, handshake errors, authentication failures, unexpected certificate warnings, and requests directed at retired hosts. Review web application firewall events, identity-provider logs, API gateway telemetry, and service-mesh metrics. Monitoring should continue through at least one normal business cycle, including overnight processing and scheduled integrations. For organisations serving customers across Australian time zones, this can expose problems that are invisible during a standard Melbourne or Sydney office-hours test.

Threat hunting should look for activity that began before the certificate was revoked. Search for unusual access by the affected service account, anomalous source locations, new administrator sessions, changes to DNS, and outbound connections from systems that held the private key. If the key was present in a developer workstation or CI/CD environment, examine adjacent repositories, build jobs, artefacts, and credentials for related exposure.

Document the final state in the incident record. Include the original certificate details, compromise assessment, CA revocation confirmation, replacement certificate details, affected services, testing results, monitoring period, stakeholder approvals, and any customer or regulator notifications. This record supports forensic analysis and demonstrates that the response addressed both the credential and the conditions that allowed it to be exposed.

Make Certificate Rotation Repeatable

An emergency replacement often reveals weaknesses in ordinary certificate management. Teams may discover that ownership is unclear, expiry alerts reach an abandoned mailbox, or no one knows whether a certificate protects a legacy service in a colocation facility. Use the post-incident review to assign accountable owners, define service criticality, and set rotation intervals that are shorter than the certificate’s maximum validity where risk justifies it.

Short-lived certificates can reduce the value of stolen credentials, especially when automated issuance and deployment are reliable. They do not remove the need for revocation because an attacker may exploit a key before expiry. Automated renewal should include health checks, staged rollout, failed-deployment alerts, and a safe mechanism for restoring service without restoring a compromised key.

Certificate pinning requires special care. It can strengthen trust in a controlled application, yet poorly managed pins can turn a routine replacement into a major outage. Where pinning is used, maintain an approved backup pin or an update path that can accept the new public key. Review partner contracts and integration runbooks so external organisations know how emergency certificate changes are communicated and verified.

The most important lesson is simple: revoke the compromised certificate, create a completely new key pair, replace every deployment, and verify the result across production, cloud, backup, partner, and legacy environments. TLS protects trust only when certificate inventory, identity controls, monitoring, and coordinated incident response work together. That is what readers should remember when a certificate compromise threatens a live service.