The hierarchical model of X.509 web certificates has long been the foundation of HTTPS authentication and encryption. A tightly guarded root certificate, whose private key is protected by physical isolation and Shamir’s secret-sharing scheme among multiple guardians, signs intermediate certificates. These in turn sign the leaf certificates actually presented by websites.
Because ordinary users cannot independently verify every root, browsers and operating systems maintain curated lists of trusted roots. Adding a custom root, such as one operated by a corporation or an individual, is possible but leaves that entity solely responsible for the security of every site it vouches for.
Russian banks have recently begun issuing certificates signed by the MinTsifry root after major Western and later Chinese certificate authorities declined to issue or renew certificates because of sanctions. The author stresses that the technical security of the MinTsifry root is comparable to other roots provided its private key remains well protected; the real question is who can authorize new subordinate certificates.
Access by intelligence services is a central concern. The text notes that the FSB is presumed to hold influence over the MinTsifry root, while the NSA is widely believed to possess similar capabilities with U.S. commercial roots. The same logic applies to other agencies such as Mossad or Iran’s KSIR.
A fundamental limitation of current X.509 implementations is the inability to require signatures from multiple independent roots for a single site. The author describes a hypothetical multi-signature model in which a site would be simultaneously validated by roots from different geopolitical spheres, making undetected man-in-the-middle attacks by any single agency far more difficult.
Until such changes are adopted, users seeking to limit exposure are advised to maintain separate browsing environments, such as a dedicated smartphone or a browser running inside a virtual machine, when accessing Russian government or banking domains.
