What actually failed
On 17 October 2026, a researcher reported that the private keys behind eight eMudhra intermediate certificates were tiny integers. Scott Helme published an independent analysis: he recovered all eight values (4, 56, 66, 115, 155, 156, 230 and 237) in under a second on a laptop, using only the public certificates. Six of the intermediates were TLS issuing CAs and two were code-signing CAs.
The browser response was fast. Chrome and Firefox blocked the certificates on 17 October. Apple followed on 19 October and Microsoft on 20 October, according to MediaNama. Mozilla's bug 2083449 records that eMudhra revoked the eight intermediates on 18 October. It also records Mozilla's emergency removal of trust in the emSign ECC Root CA - G3 on 20 October, which leaves about 30,000 certificates to migrate.
The incident has an upside. The browser trust ecosystem worked as designed: an outside researcher found the flaw, root programs distrusted the keys within hours, and so far there is no confirmed misuse. CERT-In's advisory CIAD-2026-0050 says Google found no related certificates in Certificate Transparency logs before blocking. It also says no malicious exploitation had been established as of 21 October.
The steelman for MeitY's calm
There is a fair case for the government's tone. Panic about a CA can itself do damage, and pushing the market away from Indian CAs could hurt a domestic industry. India's own digital signature and eSign infrastructure, according to the advisory, is not affected. On 21 October MeitY said some social media accounts were spreading fear and that the CCA remained 'completely unaffected', as MediaNama reports. If the goal is to avoid a trust panic over Indian certificates, measured language is reasonable.
But measured language is not the same as accountability. Calling the concern 'fear mongering' while the world's four main browser vendors were removing the CA from their trust stores sent the wrong signal. The weak keys were not a social media rumour. Mozilla's team confirmed them independently the same evening they were reported.
The explanation does not close the case
eMudhra attributes the failure to a flaw during a key ceremony on 30 April 2025, which produced a second, weak key pair that was then used by mistake. Mozilla's bug thread shows commenters asking how a certified hardware security module could have produced such keys. They are asking for ceremony records and auditor witness reports. Mozilla's bug also notes that a Mozilla questionnaire sent on 20 October with a 24-hour deadline had drawn no response by 23 October. The same bug records that eMudhra's status page dated its awareness to 18 October, a day after the report was logged.
This matters because the keys were generated in April 2025 and sat unnoticed for roughly eighteen months. Whatever audits ran in that window did not catch a key that was the number 4. A sound regime would ask whether the audit design, the auditor, or the process failed. A company-level exit does not answer that question.
Exit is not enforcement
On 22 October, eMudhra said it would leave the global SSL/TLS business. It cited social media scrutiny, rising data-sovereignty compliance costs and thin returns, and said the affected business was about 1% of revenue. It will keep its digital signature certificates and Indian signing services. The exit removes one actor from the global web PKI. It leaves the domestic rulebook unexamined.
India's framework is statutory. The Controller of Certifying Authorities licenses CAs under the IT Act, 2000. Section 30 requires each CA to use systems 'secure from intrusion and misuse' and to follow security procedures that keep electronic signatures private. A CA whose private keys can be recovered in under a second raises an obvious question under that duty. The answer should come from a public CCA review, not from the licensee's press release.
As of MediaNama's 23 October report, the CCA had issued no public statement. That silence is the real governance gap. The other licensed CAs are asked to trust that nothing similar sits in their ceremony records.
A proportionate response
The response should not be heavy-handed. Blanket licensing bans, mandatory domestic-only trust, or punitive liability would raise costs for small CAs and push them out of the market. They would also reduce the competition that gives websites alternatives. A better approach would be narrow and transparent:
- Publish a root-cause report. The CCA should release ceremony records, auditor findings and the timeline of discovery and revocation, with redactions limited to security-sensitive detail.
- Audit the other CAs on the same standard. Every licensed CA should show independent verification of key randomness, using approved random number generators and HSM controls as CERT-In's own advisory recommends.
- Align with the root programs' disclosure norms. Mozilla and Chrome expect incident reports, and a domestic regulator should expect at least as much from its licensees.
- Do not punish disclosure. Researchers who flag weak keys should be treated as part of the safety system, not as sources of 'fear'.
The lesson for open-internet policy
The web's security rests on a few hundred CAs that browsers choose to trust. That is a market-like discipline, and it worked here. National regulators add value when they fill the gaps that browser programs cannot: audits, accountability for domestic licensees and transparency. India's CCA has not yet shown that it does so. Until it explains what its oversight missed, eMudhra's exit will look less like a resolution than a quiet way to close the file.