Market Insights

eMudhra’s CA key was the number 4. Two audits missed it. Shorter certificates won’t fix that.

eMudhra is leaving the global TLS business after eight of its intermediate keys turned out to be single bytes. The industry’s answer to certificate risk is shorter lifetimes. But these intermediates were valid until 2040, and the real exposure lives in trust stores nobody inventories.

Oct 23, 2026 · 4 min read

A certificate authority, or CA, is a company your browser trusts to vouch that a website is who it says it is. Its whole job rests on one secret: a private key so large nobody can guess it.

One of eMudhra’s was 4.

What happened. Security researcher Scott Helme showed that eight eMudhra intermediate CA keys, six for website certificates and two for code signing, were single-byte numbers: 4, 56, 115, 155, 156, 230, 66 and 237. A real P-384 key should be a random 384-bit number. He recovered all eight from the public certificates in under a second (Scott Helme). Chrome and Firefox blocked them on October 17, within hours of the report (Mozilla).

On October 22, eMudhra said it is voluntarily leaving the global SSL/TLS certificate business. The keys came from a flawed step in an April 2025 key ceremony. The flaw sat undetected for 18 months and through two independent audits. Chairman V. Srinivasan said certificates were about 1% of company revenue (The Record). Google says it is “closely investigating” whether eMudhra’s remaining CA certificates stay in the Chrome Root Store.

The uncomfortable point. A business line worth about 1% of one company’s revenue held a key that could vouch for any website on the internet. That’s how web trust works. Every trusted CA is equally powerful, whatever its size. Your bank’s site is only as safe as the weakest CA your users’ devices trust.

And the audit didn’t catch it. Audits mostly confirm the process was followed and documented. Nobody checked that the output, the key itself, was actually random. Helme notes a key derived from 4 passes every structural check. The only way to find it is to try.

Why shorter certificates don’t help here. The industry’s big certificate-risk fix is shorter lifetimes. Under CA/Browser Forum ballot SC-081v3, the maximum life of a public website certificate dropped to 200 days in March 2026, goes to 100 days in March 2027 and 47 days in March 2029 (DigiCert). That’s good hygiene for the certificates on your servers.

It doesn’t touch the problem here. These were intermediates, the CAs that sign those short-lived certificates. Per Sonatype co-founder Brian Fox, quoted by The Record, the eMudhra intermediates weren’t due to expire until April 2040. Shrinking the leaf doesn’t shrink the branch.

Where the exposure really lives. Browsers fixed this fast because they pull revocation updates on their own. Lots of software doesn’t. It ships its own list of trusted CAs: Go programs, Python’s certifi package, containers, appliances, vendor tools. Fox told The Record that most organizations don’t have an inventory of where those trust stores live, and “the long tail will be older releases, forgotten containers, embedded devices.” That makes it a software supply chain problem, not a browser problem.

The market read. Certificate lifecycle management, the tools that find and renew certificates, is usually sold on the 47-day deadline. That pitch is about the certificates you issue. eMudhra shows the bigger gap is the certificates you trust: which root lists are baked into which builds, and how fast you can change them. Vendors that can answer “where does every trust store in my estate live” have the better story right now. Most CLM pitches still lead with renewal automation.

What we’d do this week. Search your builds and images for emSign roots and the affected intermediates. Mozilla’s incident page lists them. Update certifi, Go and OS trust packages in anything you ship. If you buy certificates from eMudhra, start migrating now rather than waiting for a root removal to force it. And add one line to your next CA due-diligence call: how do you test that your keys are actually random?

Sources