🛅Technology has always been sold to humans under the tag it is full-proof and will make lives easier.

⌛However, a careful study of history will tell you to proceed with caution. Every technological advancement comes at a cost and added complexity. And often this complexity soon becomes vulnerability, which has to be fixed by adding more complexity and spending more money.

🏦Banking Systems which were supposed to be the securest of all the systems have ultimately become a pawn in this cat-mouse game and have been trying to catch up with tech to their full extent, only to realize within a few decades, they are again catching up.

Before 1970s, Banks relied on physical security and basic ciphers to safeguard financial transactions. Financial records were maintained in hand-written ledgers and money was kept in secure vaults. However, there was always a possibility of someone physically accessing the records or a courier carrying sensitive financial records (e.g. transaction records, account details, etc.) could be intercepted, bribed, or attacked. Cases of forgery and tampering were quite common since bank ledgers could be accessed manually.

To counter forgery or tampering, banks started using simple encryption techniques like Caesar ciphers or monoalphabetic substitutions to encrypt financial transactions. However, these techniques were not robust as someone with knowledge of basic cryptography could easily decipher the message all along. If a bank encoded “WITHDRAW 500 GOLD” by shifting letters, an attacker intercepting it could easily reverse the shift and read the message. Some banks started using telegraphs or phone lines to communicate financial transactions, however, these two modes lacked message encryption, and lines could easily be intercepted, or telegraph messages could be altered.

Identity verification was anyways out of the question since a scammer could easily send a fake telegram pretending to be from a bank manager and instructing a branch to transfer funds to a fraudulent account.

To counter above problems, banks gradually started adopting digitization in the hope of securing financial records and transactions. ATM transactions, online banking, wire transfer, etc. all got encrypted under the assumption that even if attackers intercepted a message, they couldn’t decrypt or modify it without the secret key. However, this was never meant to be.

In 1977, The U.S. National Institute of Standards and Technology (NIST) adopted DES (Data Encryption Standard) as the first widely accepted encryption standard and it quickly became the gold standard for banking encryption, including ATMs and SWIFT (Society for Worldwide Interbank Financial Telecommunication) transactions. DES is a symmetric block cipher technique which uses 64-bit block size (amount of plain-text processed at once) with a 64-bit key for encryption and decryption. Out of the 64-bit key, 8 bits are reserved for parity checks and technically 56 bits is the key length. 56-bit key length means there are 2^56 possible combinations of keys, which is basically 72 quadrillion possible keys. This seems a lot, correct? At least to the bankers!

However, back in 1965, Gordon Moore, the co-founder of Intel, that the number of transistors on a microchip doubles roughly every two years. This meant computing will become more efficient and less expensive every two years. Not much heed was paid to the observation though. During early 70s and 80s, there were not much powerful computer chips available to commoners and hence it seemed banking systems were infallible. But in the due course, Moore’s observation and prophecy held its ground. Computing power of general available computers increased to such extent that DES encryption could be broken using brute force within hours or days.

Note: A brute-force attack is when an attacker systematically tries every possible key until the correct one is found.

First significant breakthrough was done by The Electronic Frontier Foundation (EFF) in 1998, which built a DES cracker, and it could crack a DES key within 56 hours. Speed up to available computers in 2000s, brute force decryption time decreased significantly to minutes.

Banking systems were alert not with such developments and took counter measures. Instead of increasing the DES key size, a workaround was created. New encryption technique 3DES was implemented, which encrypted data three times using three different keys. But, if one key was breakable in days or minutes, there was not much hope for such a system.

In 2001, the U.S. National Institute of Standards and Technology (NIST) selected AES (Advanced Encryption System based on Rijndael algorithm) as the replacement for DES. This system was designed to replace DES by countering its vulnerability to brute-force attacks.

While DES was based on Feistel network, where data was divided in two halves and processed through multiple rounds (3 in case of 3DES) using different subkey derived from main encryption key; AES uses a substitution-permutation network (SPN) where entire blocks of data are processed in each round and, the number of encryption rounds depend on the key length. As per the AES standards, the block size is fixed at 128 bits and allows a choice of three keys: 128 bits (AES-128), 192 bits (AES-192), and 256(AES-256) bits. For AES-128, there are ten rounds; for AES-192, there are 12 rounds; and for AES-256, there are 14 rounds.

Let us look at the possible key combinations for AES as well. AES-128 promises 2^128 possible combinations and AES-256 offers 2^256 possible combinations. An underlying statement about the capability of such a system was assumed till a year back was that Even if all supercomputers on Earth worked together, brute-forcing AES-128 or AES-256 is practically impossible.

Really?

There has been made significant progress in quantum computing over last few decades and off late top-tech companies have been unveiling their quantum chips which promise to crack the algorithms which were meant to solved in billions of years to minutes or seconds.

Quantum computing is unlike normal computing and uses qubits instead of bits. A qubit can be in state 0 or 1 or superposition of both. While normal computer can do one operation at a time, superposition means multiple possibilities can be evaluated at a time and this means exponential speed-ups for certain problems.

Some modern encryption techniques like RSA, ECC, and Diffie-Hellman encryption rely on the difficulty of factoring large numbers or discrete logarithms, while Quantum computers can run Shor’s Algorithm, which can factor large numbers exponentially faster than classical computers. This basically means an efficient quantum chip in the future will be able to break RSA-2048 in seconds or minutes. By all means public-key cryptography is at risk.

A counter question on the above point would be, does not RSA uses SHA-256 and hence, it is also safe from quantum threats.

Clarification: Yes, but SHA-256 is not what protects RSA encryption, SHA is used in RSA for digital signatures. When signing a document, RSA doesn’t encrypt the entire document (reason: RSA encryption is slow for large data), instead RSA operates on short fixed-size inputs and SHA-256 is used to hash the document by creating 256-bit fingerprint.

Example: If the message is “Hello World”, SHA-256 produces a unique fingerprint.

Hash is encrypted with its RSA private key, which in turn creates a digital signature that only the sender could have produced.

When the receiver gets the message and the signature, the signature is decrypted using the sender’s RSA public key. This will reveal the original SHA-256 hash and hash of received message can be compared against it.

Point being, SHA-256 itself is not the weak point, but the problem lies in RSA’s key structure as Quantum threats will attack RSA’s prime factorization, not SHA-256.

RSA’s security relies on the difficulty of factoring large prime numbers.

Ex: N = p x q; Shor’s algorithm can easily factor N back into p and q.
Attacker does not need to deal with SHA-256 at all.

Then, what about AES? While classical brute-force attacks on AES require 2^n operations for an n-bit key and best quantum attack on AES uses Grover’s Algorithm, which can speed up brute-force attacks, reducing the search time to 2^n/2. This means AES-128, which normally requires 2^128 operations by brute-force, would need only 2^64 quantum operations and AES-256 would require 2^128 quantum operations.

In short, AES-128 is vulnerable to quantum attacks however AES-256 is safe for the foreseeable future, until quantum chip development is also cursed by Moore’s law and within few decades raises the bar.

🗝️Ultimately Choice is ours: To what extent these costs and complexities are justifiable.