Let’s Encrypt issues certificates of which they are the CA. That cert is based on a private key generated in the server by LE's auto/certbot script.

Could a state actor MITM that transaction requesting a cert and send down a spoofed cert in it's place which would appear valid on the surface? One which the state actor could make weaker and easier to crack?

Alternatively, could a state actor hijack the process to install LE's script in the first place, supplanting it with a script that requested a cert from a CA they controlled? Again, sending back a weak cert they could crack?

Most people just blindly install and run the script. They see a cert working, and move on. They don’t validate the chain themselves. This seems a potential weak point.

I think the private key which doesn't leave the server largely protects against this, but I am not certain if that's the case, hence this question.

So how secure is are those processes? Why are the considered secure? Could a MITM attack like this work based on user carelessness to validate the chain?

  • This question makes no sense. If the relying parties already trust the Government CA, then the Government or Government CA would not need a MITM position, they can just issue a certificate under your name that'll be trusted by the RPs. – Lie Ryan Mar 12 '18 at 11:47
  • Not suggesting a Gov CA. I'm saying could a a state actor hijack the initial cert setup, intercepting the initial request and returning a "valid enough" looking cert to the LE script, or could the intercept the installation of that script in the first place causing it to request a cert from somewhere else the state actor controled. – jb510 Mar 12 '18 at 18:22

The private key never leaves the server requesting the certificate.

The public key in the certificate has to match this private key.

The certificate returned is either invalid (public key not matching private key) or it has the same security. It might be signed by another CA, but that doesn't change the security. It might be considered invalid by some browsers.

If a attacker can MITM the actual connection between client and your server, then a state actor can interfere.

  • A government CA can just issue their own CA. There's no need for them to attack the signing process. – forest Mar 12 '18 at 9:33
  • 2
    @forest as I wrote, attacking the signing process leads either to an invalid certificate or to a secure one. Other attacks are not really the scope of this question. – Josef Mar 12 '18 at 9:35
  • Yeah that's fair enough. – forest Mar 12 '18 at 9:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.