Open a website on a company laptop at many large Indian firms, click the padlock, and look at who issued the certificate. Instead of a public certificate authority, you may see something like “Corporate Inspection CA”. Your connection is still encrypted, but not end to end with the website. A device in the middle, operated by your employer, is decrypting, inspecting and re-encrypting it. This is deliberate, and understanding how it works explains a lot about modern network security and its tensions.

Proxies in two directions

A forward proxy acts on behalf of clients. Users’ browsers send requests to the proxy, and the proxy fetches from the internet on their behalf. Companies use forward proxies, increasingly delivered as cloud secure web gateways, to filter categories of sites, block known malicious destinations, scan downloads and keep logs of who visited what.

A reverse proxy acts on behalf of servers. Internet users connect to it, believing it is the website, and it forwards requests to the real application servers behind it. Load balancers, CDNs and web application firewalls are reverse proxies. They usually terminate TLS using the website’s own certificate, which is perfectly legitimate because they act for the site owner.

TLS inspection on the forward side is the more delicate case, because the proxy is decrypting traffic between your users and websites the company does not own.

How the interception works

TLS is designed to defeat anyone in the middle. A browser only accepts a certificate for a site if it is signed by an authority the browser trusts. So inspection works by adding one more trusted authority.

  1. The company creates its own internal certificate authority and pushes its root certificate to every managed laptop’s trust store through device management.
  2. A user opens a web mail site. The connection reaches the proxy, which reads the requested server name from the TLS handshake.
  3. The proxy makes its own TLS connection to the real website and validates the real certificate, as a browser would.
  4. The proxy generates, on the fly, a certificate for the same site name signed by the internal CA, and uses it to complete the handshake with the user’s browser.
  5. The browser trusts the internal CA, so it accepts the certificate. The proxy now holds two encrypted sessions and sees plain text in between, which it can scan before passing it on.

Technically, this is exactly what an attacker attempting a man-in-the-middle would like to do. The difference is that the company controls the endpoints and has installed its own trust anchor with authority to do so. On an unmanaged personal phone, without that root certificate, the same interception produces a certificate warning.

Why organisations do it

Most web traffic is encrypted, and so is most malware delivery and command traffic. Without decryption, the proxy, IPS and data loss prevention tools see only destination names and volumes. With decryption, they can scan downloaded files for malware, detect sensitive data such as customer records being uploaded to personal storage, enforce policies inside cloud applications (allowing a corporate storage tenant while blocking personal accounts on the same service), and apply intrusion signatures to payloads.

What it costs and what breaks

Performance. Decryption and re-encryption are expensive. Enabling inspection can sharply reduce the throughput of a firewall or proxy, so capacity has to be planned for it.

Applications that pin certificates. Some mobile apps, desktop clients and update services accept only a specific certificate or key, not any CA in the trust store. Inspection breaks them, and they must be added to a bypass list.

Security of the proxy itself. The proxy now sees passwords, tokens and sensitive data for the whole organisation, and its internal CA key can mint trusted certificates for any website on company devices. Both make it a high-value target. Its validation of upstream certificates also matters; a proxy that accepts invalid certificates from real sites, then re-signs them as valid, actually reduces user security. Academic studies have found exactly such weaknesses in some middlebox products.

Newer protocols. HTTP/3 over QUIC is often blocked so browsers fall back to TCP, which the proxy can handle. Encrypted Client Hello, as it spreads, hides the requested server name from passive observers, which pushes organisations toward explicit proxies or endpoint-based controls.

What to leave alone

Sensible inspection policies exclude categories where decryption adds privacy risk without much security value, typically personal banking, health and some government services. This reduces the sensitive personal data the organisation handles and limits awkward conversations about employee monitoring.

That last point has a legal dimension. Decrypting employee traffic means processing personal data. In India, the Digital Personal Data Protection Act, 2023, and its rules shape how organisations must handle such processing, and the expectations differ in other jurisdictions where staff may be based. Companies typically disclose inspection in their acceptable use policy and employment terms, limit who can access decrypted logs, and set retention periods. The details are a matter for legal counsel, not just the network team.

Inspecting less, seeing more

Some organisations are moving part of this work to the endpoint. An EDR or data protection agent on the laptop sees files and data before encryption, without intercepting TLS at all. Cloud access security brokers use SaaS providers’ APIs to inspect content already stored there. These approaches do not replace network inspection entirely, but they reduce how much traffic needs to be decrypted in the middle, and with it the performance cost, breakage and privacy exposure that come with breaking encryption on purpose.

Written by the RP Strativa Team. The CA name in the introduction is illustrative.

Leave a Reply

Trending