0

Is it possible to have an encryption algorithm/implementation that has a "backdoor", that also wouldn't introduce flaws into the encryption?

For instance, if you have a master password, once that gets leaked that person now has access to anyone's account.

Would it be possible to have TWO passwords work to decrypt: One for the authorities, and one for the user? That doesn't sound very secure, as now maybe collision are far more likely.

Then, there's also the issue of if that password given to the authorities will be secured (e.g. maybe they keep it in a excel spreadsheet in their gmail account)

2
  • 1
    No. The backdoor is a flaw in the encryption. Think about Kerchkoff's Principle: a system must remain secure, even if everything about the system is known except the key. A backdoor would be revealed by examination of the system. Commented Oct 6, 2014 at 16:11
  • 1
    This depends on your definition of "secure". If you trust big brother then backdoored encryption can be secure. If your idea of secure if that only the recipient can decrypt, then no, this is by definition impossible. The basic idea you're referring to here is Key Escrow.
    – paj28
    Commented Oct 6, 2014 at 20:53

3 Answers 3

2

To your first question, no, the backdoor is a flaw in the encryption. Think about Kerchkoff's Principle: a system must remain secure, even if everything about the system is known except the key. A backdoor would be revealed by examination of the system.

However your second clarification is talking about the concept of key escrow. I may have my secret key, but the company also has a secret key. Either one can decrypt the message. Those are common, and supported in various security packages.

2
  • Are you saying that two separate passwords decrypt two separate data blocks, or the same block?
    – RoraΖ
    Commented Oct 6, 2014 at 16:22
  • No; the way it's implemented now, it's two separate data blocks (one for each user) that are each separately encrypted and contain the master key to decrypting the message. Cryptographically, the strength is reduced by the number of extra keys. Security wise, the risk of exposure to the original message is multiplied by the number of people the keys are shared with. The original question about "would encryption that permits two passwords be secure" is difficult to answer without a specific implementation. Commented Oct 6, 2014 at 19:06
1

It isn't unheard of. Goldkey is a product that has a hierarchy of grandmaster, master, and normal.

I'm not sure this is necessarily a back door, as the intention is that a department member can't access a peer's corporate data, but a department lead can. Then usually a grandmaster key is kept in a lockbox as an insurance policy in case a master key is lost.

In the case of gold key these are physical tokens with very good security, and the whole point is for a corporate structure for where soldiers in the field may lose their keys.

0

You have asked several questions. I'll answer the first.

Yes, it is possible if the algorithm is not published. One the encryption method is known, and the only security is the secrecy of the key, it is very unlikely to devise a method that has a master key. There is probably a proof for this, but no one is interested in it since no unpublished encryption is trusted.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .