CA/PKI Weekly Digest - Oct 19 - Oct 26, 2026

The eMudhra incident turned out to be worse than it first looked. The private keys of eight eMudhra intermediates were not stolen. They were made badly: each was a number below 256, so anyone could work them out, and Mozilla says it recovered them itself within two hours of the report (bug 2083449). Others did too. Between October 20 and 21, 107,107 certificates signed with six of the leaked keys appeared in public Certificate Transparency logs. Software that checks revocation rejects them, because all of these intermediates were revoked on October 18 and 19. To protect software that does not check, Mozilla and Chrome decided on October 20 to remove trust in eMudhra's emSign ECC Root CA - G3, and eMudhra stopped issuing from every intermediate under it. eMudhra has since announced it is leaving the global TLS certificate business (press release). On October 23 Mozilla said eMudhra had not yet answered its questions about how the keys were made. Every TLS Radar scan, including the free one, flags a site whose chain runs through one of these intermediates (our write-up).

As of October 25, eMudhra's revocation and OCSP servers refuse requests from common tools such as curl and Python's standard library (bug 2085612). We checked: TLS Radar's revocation checker still gets eMudhra's lists.

GoDaddy has finished its mass revocation (bug 2083191). It corrected the total to 403,643 certificates: 378,067 were revoked and 25,576 expired first, and none was still valid on October 22, four days after the deadline (our write-up).

TrustAsia revoked 19,237 certificates issued for subdomains whose CAA records said TrustAsia must not issue for them (bug 2083471). Its system used the CAA answer for the parent domain instead of checking the subdomain itself. CAA is the DNS record that lets you list which CAs may issue for your domain, and it only protects you if the CA checks it properly.

DigiCert reported unauthorized access to one of its business systems, the one that stores information customers' representatives send to prove who they are. It says no certificate operations were affected (bug 2084460). Sectigo's own quarterly audit found that some account reviews for its CA systems were late, and that one former contractor's account was left on a retired VPN server (bug 2085275, bug 2085276).

Cybertrust Japan, which issues under SECOM's root, issued two different final certificates with the same serial number from one precertificate, because the step from CT logging to issuance ran twice (bug 2084449). A serial number is supposed to identify exactly one certificate.

The OCSP reports moved from the answers to the certificates that sign them. SHECA, Disig, IZENPE, Telia and certSIGN all reported OCSP signing certificates with missing, extra or wrongly marked fields (bug 2084097, bug 2084525, bug 2085304, bug 2085354, bug 2085356, bug 2083774, bug 2085384). One of SHECA's uses an RSA key with a public exponent of 3, far below the recommended minimum of 65,537.

The revocation-list reports keep coming too. SwissSign's and DFN's lists were valid for one second longer than allowed (bug 2083791, bug 2083715). DigiCert's list numbers went backwards at times, when they must only go up (bug 2083502). TrustAsia removed revoked certificates from its lists one update too early (bug 2085366).

Two older stories moved. Firmaprofesional finally revoked its SHA-1 intermediate, AC Firmaprofesional - CA1, on October 22, after weeks of delay (bug 2072600). Sectigo's count of active certificates covered by its disputed wording fell to 69.4 million on October 20, from 76.5 million a week earlier (bug 2069640).

A weekly summary of certificate authority incidents and compliance activity, sourced from Mozilla's public CA Program tracker.

New this week

• PKIoverheid: Delayed reporting of CCADB documentation non-compliance - source
• certSIGN: Delayed revocation on a set of test certificates - source
• IZENPE: ocsp requests filtered by user-agent - source
• NETLOCK: Outdated CP/CPS disclosure in CCADB for NetLock Arany (Class Gold) Főtanúsítvány - source
• Certigna: Serial number of a cross-signed CA certificate generated with java.util.Random - source
• eMudhra: Access to CRLs blocked for certain User-Agent headers - source
• SECOM: Multiple Final Certificates with the Same Issuer and Serial Number Issued by Cybertrust Japan - source
• Atos: OCSP Responder certificate contains subjectAltName extension - source
• D-Trust: Loss of archived RA records for S/MIME certificates at E.ON SE RA - source
• D-Trust: Published Certificate Problem Report mechanism unavailable - source
• D-Trust: Certificate name encodings inconsistent with TSPS section 7.1.4 - source
• DigiCert: Unauthorized access to business system - source
• DigiCert: Manual data entry error in CCADB - source
• SSL.com: EV JOI Subject Field Abbreviation and Registration Level Mismatch - source
• IdenTrust: Issuance of Subordinate CAs non-compliant with CP/CPS - source
• SHECA: OCSP Responder Certificate with RSA Public Exponent 3 - source
• SHECA: OCSP Responder Certificate contains certificatePolicies extension - source
• certSIGN: CRL entry with reasonCode unspecified - source
• certSIGN: OCSP with bad KeyUsage - source
• TrustAsia: Premature Removal of CRL Entries - source
• Telia: OCSP Responder Certificate does not have subject:countryName attribute - source
• IZENPE: Non-BR-Compliant OCSP Responder certificate - source
• Disig: keyUsage in OCSP responder certificate not marked critical - source
• Sectigo: Late account removal for deprecated Network Boundary Control - source
• Sectigo: Late review of accounts for a subsection of CA Systems - source
• NETLOCK: Precertificates issued by Precertificate Signing CA after 2026-03-15 - source
• SwissSign: Delayed Response to CPR - Bugzilla 2083791 - source
• PKIoverheid: Delayed Disclosure of Incident Report Beyond the Required 72-Hour Timeframe - source
• SwissSign: CRL max validity period exceeded - source

Updates on open incidents

• Firmaprofesional: Delayed revocation of SHA-1-signed AC Firmaprofesional - CA1 - source
• DigiCert: Revocation of Undisclosed Issuing CAs - source
• eMudhra: Private key compromise of Intermediate Certificates - source
• ACCV: Delayed revocation related to Bugzilla #2061746 - source
• IZENPE: revocation reason code not included in CRL issued by Root - source
• GoDaddy: Delayed Revocation of SHA-1 Signed Subordinate CA Certificates - source
• GoDaddy: Certificate Issuance Using Previously Compromised Private Key - source
• Actalis: Root CA CRL not published within 24 hours after recording revocations - source
• Actalis: Incorrect Revocation dates in CRL entries - source
• GoDaddy: Delayed Certificate Revocation - source
• FNMT: Delayed Revocation of SHA-1 Signed Subordinate CA Certificate - source
• Sectigo: No revocation after CP/CPS language involving extendedKeyUsage found to be subject to multiple interpretations - source
• Microsec: Insufficent support for "unspecified (0)" revocation reason - source
• Firmaprofesional: Chrome Root Progam Policy - Dedicated TLS hierarchy / EKU requirements - source
• Telekom Security: Failure in automatic issuance of CRLs - source
• D-TRUST: Subordinate CA Records - Outdated CPS Effective Dates in CCADB - source
• Sectigo: Certificates issued for sanctioned entities - source
• Asseco DS / Certum: Incomplete CRL Disclosure in CCADB for intermediate CAs sharing the same subjects and public keys as root CAs - source
• ACCV: OCSP responses include “archiveCutoff” in “responseExtensions” - source
• NAVER Cloud Trust Services: Subordinate CA certificates with two Reserved Certificate Policy Identifiers - source
• Actalis: Non-compliant OCSP responses and OCSP responder certificate - source
• OISTE: Specific test websites for EV not listed in CCADB - source
• eMudhra emSign PKI Services: Delegated OCSP Responder certificates containing cRLDistributionPoints extension - source
• ACCV: Issuance of Server TLS Certificates with CP/CPS Discrepancies - source
• NAVER Cloud Trust Services: Cross-certificate with incorrect accessMethod (id-ad-ocsp instead of id-ad-caIssuers) in AIA - source
• TrustAsia: CAA Mis-Issuance - source
• NETLOCK: Failure to Respond to a Certificate Problem Report Within 24 Hours - source
• TunTrust: CP/CPS markdown version missing root program attestation - source
• FNMT: Issuance of intermediates after 2019-01-01 that do not comply with Mozilla Policy - source
• CFCA: Test websites for EV roots serve OV certificates - source
• CFCA: Incorrect countryName values in OV subscriber certificates - source
• Apple: CRL is signed incorrectly - source
• D-TRUST: Reported CA key-size inconsistency in TSPS v3.1 - source
• D-Trust: Discrepancy in CRL URL Disclosures - source
• Telekom Security / DFN: CRL validity period exceeds 10 days by one second - source
• Telekom Security / DFN: crlEntryExtensions SEQUENCE of size 0 - source
• Telekom Security: Failure to respond to a certificate problem report within 24 hours - source
• SECOM: Incorrect and missing EV TLS test websites in CCADB - source
• GoDaddy: Delayed CRL Updates - source
• SSL.com: Failure to respond to Certificate Problem Report within 24 hours - source
• SSL.com: Inconsistent published CRL for one ICA - source
• Disig: CRL contains not allowed attributes - source
• DigiCert: Publishing timeline for CRL - source
• DigiCert: cRLNumber not increasing within one CRL scope - source
• DigiCert: Expired CRL - source
• Telia: CRL signature algorithm property non-conformance for EC issuer key - source
• eMudhra emSign PKI Services: InCommon RSA IGTF Server CA 4 — certificates issued with key sizes not permitted by CPS - source
• Kamu SM: Non-BR-Compliant OCSP Response Validity Interval - source
• Telekom Security: Typographical error in CPS section 7.1.6 - source
• Telekom Security: OCSP responder uses incorrect signature algorithm for P-384 signing key - source
• certSIGN: OCSP responder certificate encoded subject attributes out of order - source
• certSIGN: Missing localityName in an OV certificate - source
• Certigna: Failure to announce completion of action item within prescribed time period - source
• ANF: Potential TLS BR 7.1.2.8 violations in OCSP responder certificate - source
• Asseco DS / Certum: Incorrect AlgorithmIdentifier Encoding in OCSP responses - source
• DigiCert: Certificates issued for sanctioned entities - source
• certSIGN: Missing the countryName attribute in the Subject DN of a test DV TLS certificate - source
• ACCV: EKU Inconsistency in ACCV RSA1 TLS CA Certificates Identified During Chrome Root Inclusion Review - source
• Actalis: Issuance of Server TLS Certificates with id-kp-clientAuth against CPS - source
• iTrusChina: Reusing CRLnumber and Missing NULL Parameter of CRL signatureAlgorithm - source
• CFCA: Certificates Issued with lower case value in subject:countryName - source
• Chunghwa Telecom: Delay of 2025 WebTrust for CA Audit Reports seals for ePKI & HiPKI - source
• SwissSign: Backdating of 48+h - source
• FNMT: Late CP/CPS Update in CCADB - source
• Sectigo: Signing infrastructure migration disrupted issuance of 2 CRLs - source
• D-Trust: Missing Pre-Signing Linting for TLS Issuance - source
• Firmaprofesional: AC Firmaprofesional - CA1 excluded from audit scope - source
• Sectigo: Legacy Subordinate CA Certificate without digitalSignature KU signing OCSP responses - source
• NAVER Cloud Trust Services: OCSP responses with an empty responseExtensions - source
• TWCA: Non-critical basicConstraints in Delegated OCSP Responder Certificates - source
• SSL.com: Failure to Post all Root and Intermediate CA certificates in Repository identified in CP/CPS - source
• Consorci AOC (CATCert): EC-ACC audit statement 123 days after period end, no letter, no audit period newer than 2024-03-28 - source
• Entrust: AffirmTrust and Entrust 4K roots — newest audit period ended 2025-02-28, statement 119 days late, no letter, no newer period on record - source
• ITI (Brazil): Autoridade Certificadora Raiz Brasileira v10 — audit statement 144 days after period end, no explanatory letter, no newer audit on record - source
• MULTICERT: Standard audit statement 108 days after period end with no explanatory letter; TLS BR audit on record ended 2023-03-31 - source
• SSL.com: CRL URL typo in CRL URL disclosed to CCADB - source
• SSL.com: MPIC not performed for specific DCV path - source
• D-Trust: Missing Pre-Sign Linting for S/MIME Issuing CAs - source
• KIR: TLS certificates issued from legacy SZAFIR Trusted CA6 contrary to Chrome commitments - source
• GoDaddy: Subscriber Certificate Profiles Did Not Conform to CP/CPS - source
• Asseco DS / Certum: Delayed preliminary CPR response due to email address typo - source
• eMudhra emSign PKI Services: CP/CPS documents not updated for current requirements - source
• D-Trust: Finding #2 of 2 in 2026 ETSI Audit – Documentation of backup recovery testing - source
• D-Trust: Finding #1 of 2 in 2026 ETSI Audit – vulnerability management - source
• eMudhra emSign PKI Services: TLS certificates missing certificate-type policy OID required by CP/CPS - source
• SSL.com: Delayed disclosure of Timestamping CA in CCADB - source
• Sectigo: Violation of TLS BR 7.1.3.2.2 in OCSP responses - source
• KIR: SZAFIR ROOT CA3 TLS CRL nextUpdate exceeds 12 months - source
• SSL.com: Invalid combinations of countryName, stateOrProvinceName, and localityName attributes - source
• Firmaprofesional: certificateHold reasonCode entries in AC Firmaprofesional - CUALIFICADOS CRL - source
• Firmaprofesional: OCSP signature verification errors for Secure Web 2024 certificates - source
• Firmaprofesional: Delayed preliminary CPR response due to incomplete mailbox monitoring coverage - source
• Sectigo: Failure to announce completion of action item within prescribed time period - source
• Google Trust Services: CPS Missing root program attestation - source
• Sectigo: CP/CPS language involving extendedKeyUsage found to be subject to multiple interpretations - source
• Actalis: Undisclosed Subordinate CA Certificate - source
• GlobalSign: EV Test Certificates missing for 2 roots - source
• eMudhra emSign PKI Services: InCommon IGTF SSL CPS missing conformance statement for Baseline Requirements - source
• BEIJING CERTIFICATE AUTHORITY Co., Ltd.: OCSP responses with an empty singleExtensions - source
• Hongkong Post: CRL HTTP endpoints returned legacy media type application/x-pkcs7-crl - source
• Actalis: Incorrect CRL Distribution Point in TLS Server Certificates - source
• SECOM: Missing Prior Notification and Approval for a Cross-Certificate Extending Trust to Externally-Operated JPRS CAs - source
• SwissSign: Incorrectly encoded OCSP extension - source
• SHECA: Incorrect Disclosure of Partitioned CRLs in CCADB - source
• iTrusChina: OCSP Response ECDSA signatureAlgorithm with NULL parameters - source
• SwissSign: Root CRLs Missing Reason Code - source

Resolved this week

• Telekom Security / DFN: CRL of “DFN-Verein Certification Authority 2“ contains empty revoked certificate list - source
• IdenTrust: Expired certificates for "Revoked" test websites - source
• Let's Encrypt: Improper encoding of wildcard certificates - source
• QuoVadis: Non-BR-Compliant issuance --improper characters in DNSName (BIT sub-CA) - source
• QuoVadis: Non-BR-Compliant OCSP Responder - source
• SwissSign: Undisclosed Intermediate Certificates - source
• Certinomis / Docapost: Non-BR-Compliant OCSP Responders - source
• Telekom Security: QCStatement with http link to PDS - source
• Certinomis: Non-BR-Compliant Certificate Issuance - source
• QuoVadis: Non-BR-Compliant Certificate Issuance - source
• Amazon Trust Services: EV Root Test Web Pages Serve Non-EV (DV) Certificates - source
• PKIoverheid: TSP KPN Delayed publication of audit attestation letters in the CCADB - source
• eMudhra: Invalid CRL signatures - source
• Asseco DS / Certum: Multi-purpose Subordinate CA with missing EKU - source
• NAVER Cloud Trust Services: Missing CRL disclosure for two subordinate CAs in CCADB - source
• D-Trust: Incorrect CPS URLs in CCADB for six S/MIME CA certificate records - source
• DigiCert: EV JOI match with organizationIdentifer - source
• Sectigo: Submitted CPR not replied to within 24 hours - source