Zero-Knowledge Proof

Sometimes you need to prove something without disclosing everything. An example: you want to show at the cinema box office that you’re over 18. Normally you’d show your ID – and with it your name, address, and date of birth. A zero-knowledge proof solves this problem differently. It’s a computational procedure that lets you prove a statement is true without revealing the information behind it. The box office only learns: this person is of legal age. Nothing more. The name comes from the result: the verifier gains zero additional knowledge.

Why proving and disclosing don’t have to be the same thing

On the internet, an uncomfortable rule has applied until now: whoever wants to prove something hands over data. For a loan, you present bank statements. For an account, you scan your ID. Afterward, this data sits on someone else’s servers. If it’s stolen, the damage is permanent, because a date of birth can’t be changed.

Zero-knowledge proofs break this coupling apart. The proof is separated from the dataset. A bank could check whether your income is above a threshold without ever seeing the actual number. Privacy advocates call this data minimization: you only collect what you truly need. The EU is already discussing such methods for its planned digital identity wallet.

There’s a second benefit that has little to do with privacy. A zero-knowledge proof can often be verified much faster than the original computation takes to run. Someone who bundles a thousand transfers can generate a single short proof. Others then only need to check this one proof instead of a thousand individual transactions. That saves an enormous amount of computing time.

The principle behind proof without data

There are always two roles. The prover knows a secret, such as a password or a number. The verifier wants to be certain that this secret really exists. The prover computes a kind of receipt from the secret. This receipt only fits if they actually know the secret. Working backward from it is practically impossible.

A classic illustration is the cave with a locked door in the back passage. You claim to have the key. A friend waits at the entrance and calls out which passage you should emerge from. Without the key, you’d only guess correctly fifty percent of the time. After twenty rounds, bluffing is virtually ruled out. Yet your friend never saw the key.

In practice, this doesn’t unfold as a question-and-answer game, but in a single step. The most widespread constructions are called zk-SNARK and zk-STARK. Both produce a very short proof, often just a few hundred bytes in size. Creating it is computationally intensive, but verifying it is cheap. This very asymmetry is what makes the technology interesting.

From crypto networks to age verification

Zero-knowledge proofs are best known in the blockchain world, i.e., in publicly viewable digital ledgers. Networks like Ethereum use so-called rollups: they bundle many transactions and submit only a single proof. This significantly lowers fees. Other projects use it to fully hide amounts and senders.

Outside the crypto world, adoption is growing more slowly, but it’s noticeable. Apple and Cloudflare use similar methods to classify users as genuine humans without identifying them. The technology is also being tested for age verification on online platforms. It comes up in discussions about AI, too, when companies want to prove that a model was trained according to certain rules.

A common misconception: a zero-knowledge proof doesn’t guarantee that the input data is correct. It only proves that the computation with that data was done correctly. Anyone who feeds in a falsified figure will get a valid proof for a false claim. That’s why a trustworthy party is always needed, such as an authority that has confirmed the original data.