Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Is Cryptography?

Cryptography is the science of protecting information with mathematics: keeping it secret, proving it has not changed, and proving who produced it. Nearly every security control depends on it.

Illustration for the cybersecurity term Cryptography
Home / Blog / Cybersecurity Terms / What Is Cryptography?
17 November 2026

What Is Cryptography?

What Is Cryptography?

Cryptography is the use of mathematics to protect information in the presence of an adversary. It answers four questions: can someone else read this (confidentiality), has it been altered (integrity), did it really come from the claimed sender (authenticity), and can the sender later deny having sent it (non-repudiation). Nearly every other security control — authentication, secure communication, software updates, payment processing — rests on at least one of those guarantees.

The discipline is old. Caesar shifted letters, mediaeval diplomats used polyalphabetic ciphers, and the Second World War turned codebreaking into an industry, with the work at Bletchley Park on Enigma shortening the war by an estimated two years. What changed in the 1970s was publication: the Data Encryption Standard in 1977 and public key cryptography gave the world openly specified algorithms that anyone could attack.

Kerckhoffs’s Principle

The Dutch cryptographer Auguste Kerckhoffs wrote in 1883 that a system should remain secure even if everything about it except the key becomes public. That remains the test. Algorithms are published, standardised, and attacked for years before anyone deploys them, and all the secrecy lives in the key.

The alternative — secret algorithms, sometimes advertised as proprietary or military-grade — has failed every time it met public scrutiny. The GSM A5/1 cipher, the Mifare Classic card’s Crypto-1, and the DECT standard’s cipher were all reverse engineered and broken once researchers could see them.

Three Families of Primitive

Real protocols combine all three. A TLS 1.3 handshake agrees a shared secret with an ephemeral elliptic curve exchange, authenticates the server with a certificate signature, derives session keys with a key derivation function, and then protects the data with AES-GCM or ChaCha20-Poly1305. Each primitive does one job, and the protocol is where they are composed — which is also where things go wrong.

How Cryptography Fails

Almost never through the mathematics. The recurring causes are mundane:

  1. Key handling. Keys in source control, in container images, in configuration files next to the data they protect, or shared across every environment.
  2. Misuse of a sound primitive. ECB mode, a reused nonce, a missing integrity check, encryption without authentication.
  3. Bad randomness. Predictable values from a non-cryptographic generator, or insufficient entropy at first boot on embedded devices.
  4. Side channels. Timing, cache behaviour, and power consumption leaking key material from otherwise correct code.
  5. Obsolete algorithms left in place. MD5 in a signature path, SHA-1 in a legacy integration, TLS 1.0 on a forgotten appliance.
  6. Trusting the wrong party. Encryption protects data from everyone except whoever holds the key, and that list is often longer than the architecture diagram suggests.

The UK Regulatory View

Article 32 of UK GDPR names encryption and pseudonymisation as examples of appropriate technical measures, and the ICO’s guidance is explicit that where personal data is lost but was strongly encrypted with an uncompromised key, the risk to individuals may be low enough to change your notification obligations. That is one of the few places where a technical control has a direct regulatory payoff.

Elsewhere, PCI DSS requires strong cryptography for stored card data and for transmission over open networks; the NCSC publishes algorithm and protocol guidance for UK organisations, including its post-quantum timeline running to 2035; and financial entities under DORA must demonstrate cryptographic key management as part of ICT risk management. None of these frameworks asks whether you use cryptography. They ask whether you can show where, how, and under whose control.

What Is Changing

Two shifts are worth tracking. The first is post-quantum cryptography: NIST standardised ML-KEM, ML-DSA, and SLH-DSA in August 2024, and hybrid key exchange is already live in major browsers. The second is privacy-preserving computation — homomorphic encryption, secure multi-party computation, and zero-knowledge proofs — which allows useful work on data nobody is allowed to see, and is moving from research into finance, healthcare, and digital identity.

If you are not certain which algorithms your systems negotiate today, that is the place to begin. Our penetration test services report on cryptographic configuration alongside everything else, because weak cipher suites and mishandled keys remain among the most common findings we write up.

FAQ – Frequently Asked Questions

What Is the Difference Between Encryption and Cryptography?

Cryptography is the whole field, including hashing, signatures, key exchange, message authentication, and protocol design. Encryption is one technique within it, specifically transforming data so that only a key holder can read it. A system can use cryptography heavily without encrypting anything, for example by signing software releases or verifying message integrity.

Is Encoding the Same as Encryption?

No, and confusing them is a common and serious mistake. Base64, URL encoding, and hexadecimal change how data is represented so it survives transport, and anyone can reverse them with no key at all. If a system stores secrets in Base64 and calls it encryption, the secrets are stored in plaintext with extra steps.

Does Cryptography Protect Data Being Processed?

Traditionally no: data is decrypted in memory for the application to use it, which is why access control and monitoring remain essential. Confidential computing enclaves and homomorphic encryption aim to close that gap, the first by isolating memory in hardware and the second by computing directly on ciphertext, though both carry performance and complexity costs.

How Do I Know Which Algorithms to Use?

Follow published guidance rather than choosing independently. NCSC guidance, NIST SP 800-57, and the algorithm recommendations in your sector framework all give current parameter sizes. As a general baseline today: AES-256 or ChaCha20-Poly1305 for symmetric work, SHA-256 or SHA-3 for hashing, elliptic curves for key exchange and signatures, and a hybrid post-quantum exchange where it is available.

Related Terms

You May Be Interested In These

How the NotPetya Attack Worked

Cybersecurity Terms

17 November 2026

How the NotPetya Attack Worked

Read More
What Are SS7 Attacks?

Cybersecurity Terms

17 November 2026

What Are SS7 Attacks?

Read More
What Is a DCSync Attack?

Cybersecurity Terms

17 November 2026

What Is a DCSync Attack?

Read More