Posted in Incident Response

DigiCert's latest incident report deserves 10 minutes of every security team's time. Not because of the…

Craig M. — Chief Product Officer @ Cyberr®

DigiCert's latest incident report deserves 10 minutes of every security team's time. Not because of the data. Because of how ordinary the cause was.

From DigiCert's own filing on Mozilla's CA bug tracker:

→ In February 2025, an internal system archiving customer identity-validation data was migrated from a VPN-restricted segment to an externally reachable zone. The security review was skipped.

→ That exposed a self-registration endpoint. Accounts created through it did not get the MFA DigiCert normally requires, and their default "read-only" access could read archived data.

→ In June 2026, staff spotted non-DigiCert accounts. The review wrongly concluded they had no data access.

→ The data access (27 June to 2 July) was only confirmed in October, from retained logs and audit tables.

No certificates or CA systems were affected. But the lesson is not about PKI.

The control that failed was an assumption: "it sits behind the VPN." Migrations quietly delete that assumption while every other control stays green.

What I would check this week:

→ List the systems whose security model depends on network location, and re-verify their exposure after every migration.

→ Treat "read-only" as a privilege. New accounts should start with no access.

→ Put self-registration and account-creation endpoints in your attack surface monitoring, and enforce MFA at the identity provider rather than app by app.

→ When triage says "no impact", prove it from logs. And keep those logs long enough to revisit the call.

Credit to DigiCert for the transparency.

How do you stop security reviews being skipped when a migration is up against a deadline?

#IncidentResponse #CloudSecurity #IdentitySecurity

8 reactions · 7 comments

Topics: #identitysecurity, #cloudsecurity, #incidentresponse

Comments

Schomer: The moment a control's validation model relies on an IP block or a VPN lease, it introduces a dynamic variable that standard posture scanners can miss during a crunch-time migration. Deadlines create a forcing function where velocity outpaces verification. The only way to stop security reviews from being skipped is to make the boundary enforcement deterministic rather than procedural. If the ident

Victor Font: Craig, the critical governance failure here is that a change in infrastructure invalidated an existing security assumption without triggering a corresponding reassessment of risk. This is precisely where change management must become an enforceable control rather than a procedural expectation. Security reviews should be mandatory release gates for migrations that alter network exposure, identity b

Victor Font: Schomer Schomer, I agree with the architectural principle. Security boundaries should remain enforceable regardless of changes in network topology, and identity-based controls are far more resilient than assumptions about network location. I would add one important distinction: deterministic enforcement does not eliminate the need for governance; it changes what governance must verify. Someone sti

Schomer: Victor, excellent distinction. You’ve hit on the exact nexus where deterministic engineering meets defensible assurance. To your point, moving enforcement to Layer Zero doesn’t make governance obsolete; it shifts the governance mandate from a procedural checklist to continuous cryptographic and structural verification. When enforcement is hard-coded into the entry threshold as an invariant protoco

Victor Font: Schomer Schomer, I think we are converging on an important architectural principle: governance becomes substantially more effective when policy requirements are translated into continuously enforceable technical constraints. That is a meaningful advancement over relying exclusively on procedural compliance. I would qualify one point, however. An immutable telemetry stream establishes what the enfo

Read and join the discussion on Cyberr