Codecademy Team

Cryptographic Attacks

Learn about attacks and techniques that bypass cryptography.

What we’ll be learning

Cryptography is a crucial pillar of modern security, so it’s no wonder hackers have spent so much time thinking about how to bypass it. Cryptographic attacks are attacks that attempt to bypass security measures that rely on cryptography. Some of these attacks target the implementation of cryptography, while others take advantage of the math itself.

Exploiting the math

Hash collisions

If you have 5 pigeons and 4 pigeonholes for them to nest in, at least two of the pigeons are going to have to share. This is an example of the Pigeonhole Principle, a simple but powerful mathematical principle that says if you have N “pigeonholes” and more than N “pigeons”, you can’t give each pigeon its own hole.

An image of three pigeons looking at two pigeon houses. Someone out of view is saying "You're going to have to learn to share."

Instead of pigeons and pigeonholes, let’s think about hashing algorithms: Hashing algorithms have fixed-length outputs. SHA-256, for example, always produces a 256-bit output. However, hashing algorithms don’t generally have limits on the size of their limits, at least practically speaking. What this means is that there are more inputs (pigeons) than there are outputs (pigeonholes), and therefore there must be outputs that can result from more than one input. When a hashing algorithm produces the same output for two separate inputs, this is known as a hash collision.

In general, the larger the output size of a hashing algorithm, the harder it is to find hash collisions in that algorithm.

Birthday attacks

Imagine you’re looking for a hash collision. If you’re looking for a hash collision that produces a specific output, it’s going to be harder to find than a hash collision that produces any output. A birthday attack is a method of forging digital signatures that exploit this property of hash collisions. If you have a legitimate file, and you’re trying to find a malicious file with the same hash, it’s going to be easier if you have multiple, subtly different versions of the legitimate file that produce different hashes.

Birthday attacks are named after the birthday paradox, in which the odds are 50% that at least one pair of people in a group of 23 will have the same birthday.

Exploiting the implementation

A door is only as strong as the frame that it’s mounted on. While some types of cryptography can be attacked mathematically, it’s often much easier to just “go around” the cryptographic security altogether.

A drawing of a wall with a really secure-looking door next to a broken window. The door is secured with a padlock labeled “extra-strength cryptography”.

SSL stripping

HTTPS is a version of the HTTP protocol that is secured by SSL. A clever attacker can downgrade a connection from HTTPS to insecure HTTP, in what is known as SSL stripping. This allows an attacker to bypass the security implemented by HTTPS, such as verifying that a website is a website it claims to be.

SSL stripping is fairly easy to prevent, by configuring browsers and servers to refuse standard HTTP connections and only use HTTPS. The HTTP Strict Transport Security (HSTS) header is one tool that servers can use to prevent SSL stripping.

Session replay attacks

Session replay attacks are where an attacker “replays” stolen information. For example, if an attacker is able to intercept a communication where you log into your bank, they might be able to resend the same information and get into your bank account, even if they didn’t know your credentials.

Preventing session replay attacks can involve timestamping requests, identifying suspicious information in a request, or ensuring that each request is unique in some way, and is only valid once.

Pass the hash attacks

Properly stored passwords are stored as salted hashes and are very difficult to crack. To check if an entered password is correct, it is run through the same process, and compared to the hash for the correct password. If they match, the password is correct… But what if an attack is able to steal the correct hash?

Most systems won’t be vulnerable to simply entering the correct hash in the password field, since it would get hashed again and produce a different result. However, some systems do accept password hashes as authentication credentials. This is known as a pass the hash attack and is often used to log onto other computers within a compromised network.

Another example of this type of vulnerability would be if a website failed to use a secure connection for authentication, and hashed the user’s password client-side. An attacker would be able to intercept the hash from the unencrypted traffic and use the same hash to log in at their leisure. Unfortunately, this is not a fictional example.

Pass the hash attacks are only possible if an attacker is able to obtain the hash. If an attacker has local administrative access on a computer, for example, they would be able to steal the hashes for user accounts on that computer, and potentially use them to log into other computers. For this scenario, good defenses include password management/vaulting systems and ensuring that the principle of least privilege is followed for all accounts.

Initialization vectors

An initialization vector (IV), also known as a nonce (“number only used once”), is a value that sets the initial state of a cryptographic algorithm. In a well-designed system, the IV will be difficult to predict, since a predictable IV means a predictable output from the cryptographic algorithm.

Unfortunately, not all systems are well-designed. An initialization vector attack is where an attacker is able to predict the IV used for encryption, allowing them to defeat the encryption and access supposedly secure data. A poorly implemented IV might start from the same value each time the machine is powered on, or use a badly designed pseudo-random number generator that allows an attacker to predict future IVs from past IVs.

Conclusion

Cryptography is a powerful tool for security, but it’s not a universal solution. Like any security measure, it can be bypassed if it is not properly implemented or utilized, and can’t provide perfect security.