TL;DR
- On October 19, 2026, eMudhra's eight intermediate private keys were found to be extremely weak—some as simple as single digits (4, 56, 66, 115, 155, 156, 230, 237)—and could be recovered instantly from public
- The weak keys enabled attackers to forge certificates for any domain that clients not enforcing Certificate Transparency would accept, a fundamental difference from previous CA compromises where key access was contained
- Evidence shows forged certificates were created within hours of the weakness becoming public, including certificates with invalid SANs (Subject Alternative Names) longer than public CAs permit
- One week after disclosure, at least half of probes across large networks still accepted the compromised chain, indicating slow patching adoption outside modern browser environments
- Weak keys discovered
- October 19, 2026
- Previous CA incidents
- DigiNotar, TÜRKTRUST, TrustWave
- Forged cert example
- October 20, 18:19 UTC (within 1
What happened
On October 19, 2026, eMudhra, a certificate authority, revealed that eight of its intermediate private keys had been generated with extremely weak entropy. The keys were trivial numbers: 4, 56, 66, 115, 155, 156, 230, and 237—when they should have been 115-digit random numbers. eMudhra has announced it is exiting the public CA business and attributed the failure to a mistake in the key generation script, which it has not yet shared. The organization signed a certificate revocation list (CRL) with what may have been a correctly generated key, then corrected it using one of the weak keys.
Why this is unprecedented
Unlike past certificate authority compromises such as DigiNotar, TÜRKTRUST, and TrustWave—where key access was constrained to a specific set of attackers or organizations—the eMudhra keys could be recovered almost instantly by anyone from the public certificates themselves. This means any attacker or script with the public certificates could independently derive the private signing keys and issue fraudulent certificates for any domain. An unpatched client that does not enforce Certificate Transparency would accept these forged certificates as legitimate.
Exploitation evidence
Evidence of active exploitation appeared within hours of public disclosure. Cloudflare observed approximately 22,000 attempted handshakes using forged eMudhra certificates in the week following the breach. About two-thirds of attempts targeted test domains such as emudhra.we-have-a-problem.com, which Cloudflare had referenced in guidance. Most of the remaining third involved origins that appeared to have created self-signed certificates using the weak intermediate.
Among the remaining handshakes, roughly 160 stood out: the certificates contained Subject Alternative Names (SANs) for unrelated organizations' domains, including some consumer IoT device domains. These forged certificates were issued after the keys leaked, and some were valid for over two years—longer than public CAs are permitted to issue. Notably, one forged certificate was dated October 20 at 18:19 UTC, less than an hour after Mozilla's bug report became public. However, these certificates appeared to be server defaults rather than active interception attempts, since they named domains unrelated to the server name indication (SNI) requested and would have failed hostname verification. Cloudflare found no evidence that the weak keys were used to successfully intercept its own connections.
Cloudflare's response and current patching status
Cloudflare began patching high-impact systems on October 19. Systems responsible for outbound connections, including customer origin-facing proxies using Pingora, were patched within hours to reject the vulnerable eMudhra keys. An AI-assisted source code analysis completed within four hours identified all systems trusting these keys, and high-impact patches were deployed immediately.
To measure visitor vulnerability without breaking connections, Cloudflare rolled out probe connections on challenge pages: one probe to a domain with an eMudhra-signed certificate and a control probe to a regularly signed domain. Analysis of 50 million probes revealed a concerning picture. Modern, up-to-date browsers are mostly patched, but bots, older browsers, and embedded browsers such as Android WebView lag significantly, with 30-40% of Android WebView traffic still accepting the compromised chain. Across large networks, more than half of probes still accept the compromised chain. After an initial weekend of patching activity following disclosure, adoption appears largely flat, suggesting most systems remain vulnerable.
What to do now
Developers and system administrators should prioritize patching their systems and clients. Patch instructions are available on a crowdsourced website maintained by Dennis Jackson, with particular guidance for developers on macOS. Cloudflare customers using WARP or Tunnel should consult the company's earlier advisory for specific actions required for those products. Going forward, Cloudflare is exploring projects like upki that enable certificate trust stores to be dynamically updated with revocation information and Certificate Transparency keys, rather than relying on static lists of root certificate authorities.
Industry implications
The eMudhra incident exposes gaps in certificate authority oversight, including the role of auditors and cross-signers in detecting such failures. Multiple open questions remain about how weak keys passed through existing controls. The slow patching adoption observed a week after disclosure underscores the challenge of coordinating security updates across diverse systems, particularly non-browser software and embedded clients. As vulnerabilities are discovered at an accelerating pace, the ability to revoke and distribute trust updates quickly is becoming critical to Internet security.
Check before upgrading
- eMudhra is exiting the public CA business following the disclosure of the weak key generation failure
Source
Cloudflare Bloghttps://blog.cloudflare.com/emudhra-patching/AI-assisted, verified against the source. This post was drafted with AI from the primary source linked above and checked automatically against it. For anything you depend on, read the original.
- #certificateauthority
- #cryptography
- #pki
- #incidentresponse
- #vulnerability


