The Morning After We Pull a Root of Trust, Nobody Owns It
OPINION
In June 2024, Google’s Chrome Root Program said it would stop trusting new Transport Layer Security (TLS) certificates from Entrust. Behind the decision, years of compliance failures and a clear technical call. The decision was right.
The fallout became someone else’s responsibility.
That is the part we keep getting wrong. We are good at the technical decision to remove a trust anchor. Root programs at Chrome, Mozilla, Microsoft, and Apple make that call well. What we lack is a way to coordinate what happens the morning after. Trust continuity is a national readiness problem hiding inside a browser setting.
Web PKI appears distributed from the outside. It is not. A small set of embedded roots underwrites TLS, code signing, S/MIME, and the machine-to-machine auth that runs the economy. Pull one, and the blast radius is not one website. Every service is chained to it.
The record is direct. DigiNotar in 2011, breached more than 500 fraudulent certificates; the certificate authority (CA) did not survive. Symantec in 2017 wound down after years of misuse. TrustCor in 2022. Entrust in 2024. Every one was handled. No one forced a coordinated national response, because it was unnecessary. The pain stayed inside IT teams, which swapped a certificate before customers noticed.
That is the trap. We read four clean recoveries and think the issue is solved. It isn’t. It just hasn’t been tested at scale.
Picture the scaled version. A regional bank’s only CA is distrusted. It cannot process transactions or authenticate systems. Its team scrambles, but the CA is overwhelmed. Within hours, customers are locked out and regulators want answers. Competitors who issued from a second CA do not miss a beat.
And the conditions that made the past events survivable are eroding. Two forces are changing the math.
Cryptography & AI Alter the Situation
First, the cryptography under every trust anchor is on a clock. The National Institute of Standards and Technology (NIST) has standardized the post-quantum replacements, FIPS 203, 204, and 205. That is a forced migration of the primitives that sign and secure everything, on a schedule you do not control.
Second, AI lowers the attacker’s cost to find weak keys, scale social engineering against CA staff, and probe signing pipelines for the one gap. Rare events get more plausible. Planned migrations get interrupted by sudden ones.
Here is the uncomfortable part. No one owns the morning after. The Cybersecurity and Infrastructure Security Agency (CISA) coordinates cyber incidents but does not own the trust decision. The CA/Browser Forum sets issuance rules, not national response. NIST publishes guidance. The Federal PKI governs the government’s own certificates. Each owns a slice. None owns the cross-sector cleanup. Today, people about to be affected have no reliable channel to learn that a major CA will be distrusted before the public does.
The issuers have started to move. In July 2025, the CA/Browser Forum passed Ballot SC-089, which now requires every publicly trusted TLS CA to maintain and annually test a mass revocation plan. That is the right idea, made mandatory. But it binds the issuers, not the rest of us. The enterprises and sectors that have to absorb a mass revocation still have no matching duty to plan, test, or align.
We should treat trust continuity the same way we treat electric-grid black-start or DNS recovery. Those plans are named and rehearsed long before the bad day. Public-key trust deserves the same standing.
Two things have to be true. Someone should coordinate at the national level, not a new agency whose only job is to connect the people making the distrust call with the sectors who live with it. And the playbooks must be in place before the event. Emergency root removal, intermediate CA compromise, a stolen code-signing key, a forced algorithm sunset, a cross-sector cascade: Each can be written and tested while everyone is calm.
You do not need to wait for any of that. Make three moves this quarter.
-
Build a certificate and key inventory. You can only rotate what you can see. It is the highest-value move most teams are still missing.
-
Name your liaison. One owner must run trust continuity and have the authority to pull people in.
-
Run the tabletop. «Our primary CA gets distrusted in 30 days.» Walk it end-to-end and write down where it breaks. Then run it again against the post-quantum migration, because that one is already on the calendar. And issue from more than one CA, so a distrust event is a switch, not a rebuild.
The technical decision is not the problem. The morning after is ours to own. When the next root is pulled, the silence will be the sound of your sector scrambling. Break it now.
Este contenido de ciberseguridad nos ha llegado de: DarkReading
Puedes encontrar la noticia original en el siguiente enlace: https://www.darkreading.com/cyber-risk/morning-after-we-pull-root-of-trust-nobody-owns-it
