First screen, captured 5 days ago.

eMudhra intermediate CA compromise

Disclosure page reporting that private keys of eight eMudhra intermediate CA certificates trusted by major root programs were compromised, with signed proof.

What HN thinks

A CA picked a private key of 4, and the thread is still deciding whether to laugh, cry, or cite xkcd 221 for the ninth time.

Gross negligence, distrust eMudhra
Single-byte private keys, an off-script key ceremony, and re-signing CRLs with the bad keys when the good ones failed show a process failure severe enough that the CA shouldn't be trusted again.
CT doesn't protect non-browsers
Browsers enforce Certificate Transparency, so nothing went unnoticed there. But curl, pip, updaters and other clients don't check SCTs, so anyone holding the keys could quietly MITM them.
Revocation outside browsers is broken
Firefox, Chrome, Apple and Microsoft pushed blocks quickly through OneCRL-style lists. The rest of the ecosystem has little working revocation, and Firefox on Android was slow to pick up the block.

Consensus: Everyone agrees this was an inexcusable key-generation blunder rather than an attack. Commenters also agree that browser users were protected by CT and fast blocklists, while non-browser TLS clients remain exposed.

Recommended in the comments

upki ↗ (library): Suggested as a way to give non-browser software the revocation tooling that browsers already have.
CRLite ↗ (tool): Cited as modern revocation done at scale, though commenters noted it only covers end-entity certificates, not intermediates.
OneCRL ↗ (service): It let Mozilla block the affected intermediates within hours, without users needing a browser update.
ctlogs.dev ↗ (service): Used to confirm that nothing issued by the affected intermediates appeared in CT logs before the block.

Summarised by Claude from the comment thread 3 days ago.

Points: 1124
Comments: 503
Posted: 2026-10-20 by galadran
Category: Security

emudhra certificate-authority pki tls key-compromise web-pki

All captures:

Looks similar

Screenshot of danielmangum.com: Your MCU likely escaped the eMudhra intermediate CA compromise

Your MCU likely escaped the eMudhra intermediate CA compromise (danielmangum.com)

47 points by hasheddan 4 days ago | 28 comments

Blog post explaining why microcontroller-based IoT devices are likely unaffected by the eMudhra intermediate CA private key compromise, due to minimal pinned root CA bundles.

emudhra certificate-authority web-pki tls

Screenshot of scotthelme.co.uk: Eight Bits of Trust: eMudhra's CA keys cracked in milliseconds

Eight Bits of Trust: eMudhra's CA keys cracked in milliseconds (scotthelme.co.uk)

875 points by hkc 5 days ago | 456 comments

Scott Helme blog post explaining how eight eMudhra intermediate CA private keys were single-byte P-384 values recoverable in milliseconds, and who remains exposed.

emudhra certificate-authorities pki tls

Screenshot of dadrian.io: How to distrust a CA without any certificate errors

Looks similar

Your MCU likely escaped the eMudhra intermediate CA compromise (danielmangum.com)

Eight Bits of Trust: eMudhra's CA keys cracked in milliseconds (scotthelme.co.uk)

How to distrust a CA without any certificate errors (dadrian.io)

Okta – Username Above 52 Characters Security Advisory (trust.okta.com)

Use KeePassXC to sign your Git commits (code.mendhak.com)

KeePassXC Audit Report (keepassxc.org)

Related stories

Encrypting private data and private communications is now an ethical duty (blog.tripu.info)

eMudhra Login Bypass (mastodon.catgirl.cloud)

Claude Cowork exfiltrates files (promptarmor.com)

I've factored the RSA keys of a Certificate Authority from the 90s (mcpherrin.ca)

PeerTube is a free, decentralized and federated video platform (github.com)

Xray-core concealed a certificate verification bypass vulnerability (github.com)