Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Introduction
A cryptographically secure RNG is not just another random number generator. It is a specialized tool designed to withstand attack. Standard pseudo-random number generators work fine for games and simulations, but they have a fatal flaw for security. Anyone who discovers the algorithm’s seed can predict every number that follows. In cryptography, predictability means catastrophe.
This guide explains what makes a random number generator cryptographically secure, how it differs from everyday RNGs, and where you encounter these systems daily. For the foundational difference between random number generator types, see our true vs. pseudo RNG guide . For a broad overview of all applications, read our pillar post on random number generators .
What Makes an RNG Cryptographically Secure?
A cryptographically secure RNG must satisfy two strict requirements. First, it must pass the next-bit test. Given all the previous bits of output, an attacker should have no better than a 50% chance of correctly guessing the next bit. There is no pattern to exploit. Second, it must withstand state compromise extension attacks. Even if an attacker learns the generator’s internal state at a specific moment, they cannot work backward to determine previous outputs. Forward secrecy is not optional.
Standard pseudo-random generators meet neither requirement. Their output is statistically random but mathematically predictable. A secure generator adds entropy from unpredictable sources—keystroke timing, mouse movements, network packet arrival times, or dedicated hardware noise—to continuously refresh its internal state.
Where CSPRNGs Protect You Daily
You encounter a cryptographically secure RNG every day, often without realizing it. When your browser establishes an HTTPS connection to a website, a CSPRNG generates the temporary session keys that encrypt your data. When you log into an account, the authentication tokens that keep you signed in rely on secure random numbers. When you generate a new password, the random character sequence should come from a secure source.
Even blockchain and cryptocurrency systems depend on secure randomness. The private keys that protect cryptocurrency wallets are simply very large random numbers. If an attacker can predict the RNG that generated those keys, they can steal every asset the wallet holds.
For online gaming with real-money stakes, many jurisdictions now require casinos and poker sites to use audited CSPRNGs. This ensures that neither the player nor the house can manipulate outcomes. Our RNG for gaming guide explains these fairness requirements in detail.
The Entropy Problem
A cryptographically secure RNG is only as good as its entropy source. Computers are deterministic machines—they struggle to produce true unpredictability on their own. When a server boots up, it must gather enough random noise from its environment before it can securely generate keys. This period is called entropy collection.
Some systems use dedicated hardware random number generators that sample electrical noise or quantum effects. Intel and AMD processors include built-in instruction sets like RDRAND that access on-chip entropy sources. Cloud providers offer virtualized entropy services for servers running in data centers. Each approach aims to solve the same problem: giving deterministic machines access to genuine unpredictability.
For researchers running simulations that require reproducibility, standard pseudo-random generators with documented seeds work well. Our RNG for statistics guide explains how to use seeds properly. For security applications, however, a reproducible seed is a vulnerability, not a feature.
Never Roll Your Own Crypto
The most important rule in cryptography applies directly to random number generation: never build your own. A cryptographically secure RNG is subtle and easy to get wrong. Even major companies have shipped products with broken random number generators, leaving millions of users exposed.
Instead, use vetted implementations. On Unix systems, read from /dev/urandom. On Windows, use the CryptGenRandom API. In Python, use the secrets module. These tools come from decades of review and battle-testing. They handle entropy collection, state management, and all the edge cases that trip up custom implementations.
For a curated list of tools that include secure generation features, see our best online RNG tools guide .
Conclusion
A cryptographically secure RNG is the foundation of digital security. Without unpredictable random numbers, encryption keys become guessable, authentication tokens become forgeable, and private data becomes public. Standard pseudo-random generators serve well for games and simulations, but they have no place in security contexts. Use vetted, audited secure generators, respect the entropy your system provides, and never roll your own.
