In a world without privacy, we create it.

Veldrinby CounterCustody
Download on the App StoreDownload on the App Store
Scan to download VELDRIN on the App Store.
TL;DRWant the short version? Click here.
The problem
Private files sit in a camera roll anyone can scroll, an album anyone can open, or a cloud you don't control.
The fix
Veldrin keeps them in a pattern-locked vault on your iPhone. No account, no vault list, and a red emergency button for when you're pressed to open it.
In short
Zero-knowledge encryption means the provider never holds your keys and cannot read your data.

Why you need this

Plenty of products say your data is private, and few explain who actually holds the key. That gap is where trust is lost, usually at the worst moment. Veldrin keeps your files in an encrypted vault on your device, with no account to hand anyone, so there is no company login standing between you and your own pictures. Privacy you can reason about is worth more than a promise on a label.

Guide · 10 min read

What is zero-knowledge encryption?

Zero-knowledge encryption is a label vendors use for a design in which your data is encrypted with a key the provider never has, so the provider can store it but can't read it. The price is that if you lose the key, nobody can get your data back.

How does zero-knowledge encryption work, in one example?

Suppose a service stores your files. In the ordinary arrangement the service encrypts them on its servers and keeps the key, which protects against someone walking out with a disk. It does nothing about the service itself, a staff member with access, or an attacker who breaches the company and finds the keys sitting beside the data.

Encryption happens on the device first, so what reaches iCloud is ciphertext.

In a zero-knowledge design the key is made on your device from something only you hold, such as a password or a recovery key, and the files are encrypted before they leave. The service ends up holding ciphertext it can't open. The phrase refers to what the provider knows about your content, which is meant to be nothing.

It isn't a formal standard. In cryptography, a "zero-knowledge proof" is a different and narrower idea, about proving you know something without revealing it. When a storage or backup app says "zero-knowledge encryption" it almost always means provider-blind, client-side encryption, and the sensible follow-up is to ask the company what it means by the term.

How does it differ from the other kinds of encryption?

TermWho holds the keyCan the provider read the data?
Encryption in transitNobody stores a key; it protects the connectionYes, once it arrives
Encryption at rest (provider-held keys)The providerYes, and so can anyone who breaks in
End-to-end encryptionOnly the people at the endsNo, for the data that is end-to-end encrypted
Zero-knowledge (as storage vendors use it)Only youNo, for the content they say it covers

End-to-end encryption and zero-knowledge storage overlap heavily without being the same thing. End-to-end usually describes messages between people, while zero-knowledge usually describes a service that stores your data on your behalf. Neither word guarantees that a product does what it says; that depends on the details, which is where the rest of this page spends its time.

What does Apple's iCloud show about the difference?

Apple's iCloud data security overview shows both approaches inside one company. By default, many iCloud categories are encrypted but Apple holds the keys, a group that includes iCloud Backup, iCloud Drive, Photos and Notes. A smaller set, including Passwords and Keychain, Health and Messages in iCloud, is end-to-end encrypted by default, with the keys on your trusted devices. If you turn on the optional Advanced Data Protection, the end-to-end list grows to include backups, Drive, Photos and Notes, and Apple says it then has none of the keys. Mail, Contacts and Calendars stay Apple-held.

The same page describes the trade. With default protection, Apple can help you recover data it holds the keys for if you lose access. With Advanced Data Protection it can't, and you rely on your device passcode, a recovery contact or a recovery key. That's what zero-knowledge costs in practice.

What does zero-knowledge encryption protect, and what does it leave open?

  • It protects the content, meaning the file or message body, from the provider and from anyone who breaks into the provider's storage.
  • It does not protect you from your own device. If someone unlocks your phone, or malware sits on it, the content is readable there.
  • It does not hide metadata by default. The provider may still know that you have an account, when you connect, roughly how much you store and with whom you share. Ask what is and isn't encrypted.
  • It does not cover everything a company stores. Search indexes, thumbnails, previews and account details may sit outside the encrypted part.
  • It does not rescue a weak password. The key comes from something you hold, so a guessable password makes the encryption only as strong as the guess.

Why do people want the provider to be blind?

There are three reasons, and they stack. A breach of the provider exposes ciphertext and not content. The provider can't hand over what it can't read, which limits what a thief with stolen credentials can obtain, although account records and anything the encryption doesn't cover stay exposed. And you no longer have to take "we won't look" on trust, because zero-knowledge moves that policy into the system itself, where nobody has to keep honoring it.

Breaches are routine, so assume the server will be opened

Large collections of personal records leak with depressing regularity, and the usual cause is mundane: a misconfigured storage bucket, a stolen employee credential, an unpatched server. A company that holds readable data has to win every day, forever. A company holding only ciphertext can lose the whole server and hand the attacker a heap of noise, provided the keys really were never there. When you read about a breach, ask what the attacker could read. The answer separates designs more cleanly than any promise about security culture.

Stolen credentials reach what the provider can actually produce

An intruder inside a company can take whatever the company holds and can read. A provider without your key holds ciphertext, account records and logs, and that's the end of it, since it can't decrypt on request. This doesn't protect you from someone who gets your unlocked device or your key. It changes who can be tricked or breached, and it moves the weak point from the company to the person who holds the key.

"Trust us" is a policy, and policies change

Privacy policies get rewritten, companies get acquired, employees leave with access, and a promise made by one management team binds nobody after the next one arrives. A design where the provider can't read your files doesn't depend on anyone's good behavior in 2030. That's the case for provider-blind encryption in one sentence, and it's also the reason to be suspicious of a product that asks for your trust without saying how the keys are arranged.

What do you give up?

Recovery, mostly. If the key comes only from your password, no reset link exists that can bring your files back, and the operator on the other end of a support chat can only tell you so. Good products offer a recovery key, a recovery contact or a second device; each is a copy of the means to open your data, and each needs protecting as carefully as the original. Features that need the server to read content, such as full-text search across the cloud or previews, get harder. Sharing needs extra design, because every new person needs access to the key.

Which standards sit underneath, and where does NIST come in?

Zero-knowledge describes who holds the keys, and the strength of the encryption itself comes from published standards. In the United States the National Institute of Standards and Technology publishes them: FIPS 197 defines the Advanced Encryption Standard, a series of NIST special publications describes how block ciphers are used safely in modes of operation, and another covers deriving keys from passwords. Cryptographers around the world use these same primitives, which is useful, because they have been attacked in public for decades. NIST also runs a validation program that tests specific cryptographic modules against FIPS 140, and a validated module is a statement about an implementation, not about the product wrapped around it.

The practical lesson is that a good standard can't rescue a bad arrangement. A product can use an excellent cipher and still keep the key on its own server, in which case the cipher defends nothing that matters here. Conversely, a clear description of how the keys are made, stored and used says more than a list of algorithm names. Our guide to AES-256 encryption explains what the cipher does and what it leaves to the rest of the design.

Does zero-knowledge encryption make an app slower?

On a modern phone or laptop, encrypting and decrypting files is fast enough that you won't notice it for photos, documents and ordinary video. The delays people do notice come from elsewhere. Turning a password into a key is deliberately slow, so that guessing is expensive, and it happens once at unlock. Uploading encrypted files takes as long as uploading any files of the same size. The real cost is functional: anything that depends on the server reading your content, such as searching inside documents or generating previews, has to happen on your device or not at all.

How can you test a vendor's zero-knowledge claim?

  1. Find the specific claim. Does the app say what's encrypted, on which device, with which key, and who can hold it?
  2. Ask where the key comes from. If the answer is "we manage it for you", it isn't zero-knowledge.
  3. Look for the recovery story. A product that says nobody can recover your data and also says it can reset your password is contradicting itself.
  4. Check the limits page. A careful product tells you what isn't covered: metadata, shared items, backups, and what happens on a stolen unlocked device.
  5. Look for outside review. A published technical description and an independent audit are signs of care. Without them the claim stays unproven.
  6. Notice the verbs. Wording that describes a goal says nothing about what the product does, and a flat "is" needs evidence.

Our guide to end-to-end encryption for photos runs through the same questions for photo storage, and iCloud photos encryption covers Apple's side in more detail. Our manifesto lists the principles behind how we describe our own product.

How does Veldrin arrange its keys?

We avoid the label for our own product and describe the mechanism, since the label means different things to different vendors. Veldrin's design has three layers. The pattern you draw on the grid of dots derives a key. That key opens an encrypted index, and the index holds a separate random master key that protects the file contents. Because the pattern guards a wrapper around the file key, changing the pattern re-wraps one key without re-encrypting every file you own.

There is no Veldrin account and no password field, so a vault opens with the pattern or the recovery phrase. If you start a backup, the vault is encrypted on the device before it is saved to a place you choose. Restoring needs your pattern or your nine-word recovery phrase. Limits: a backup is only as safe as the place you save it, and a person who is looking at an unlocked phone sees what you see. Our account-free photo vault guide goes through how first run works without sign-up.

Summary: what to remember

Zero-knowledge, as storage vendors use it, means the provider holds ciphertext and never the key, so a breach reaches less. You pay for that with recovery, because nobody can reset a key they never had, so keep a recovery phrase or recovery key. It protects content, leaves metadata exposed, and does nothing about an unlocked phone. Test any claim by asking where the key is made, who can hold a copy, what the recovery path is and what's outside the encryption.

Questions people ask

How is zero-knowledge encryption different from end-to-end encryption?

They share a lot of ground. End-to-end usually describes data moving between people, and zero-knowledge usually describes a service that stores your data and cannot read it. Both depend on who holds the keys.

What if I forget the password to a zero-knowledge service?

If the key depends only on your password, the provider can't recover your data. Better products let you set up a recovery key or recovery contact in advance, and those need safekeeping too.

Can a hacker get my data from a zero-knowledge service?

A provider without the key can't decrypt the content, so a breach doesn't expose it. The provider may still hold account records, metadata and anything the encryption doesn't cover.

Does a zero-knowledge app collect nothing about me?

No. The term describes whether the provider can read your content. It says nothing about analytics, account details, billing records or metadata, so read the privacy policy separately.

Is zero-knowledge encryption secure against everything?

No. It protects content from the provider and from anyone who breaches the provider. It doesn't stop malware on your device, a weak password, someone who sees your unlocked screen, or metadata the service still records.

How do I know if an app really uses zero-knowledge encryption?

Look for a written description of where the key is created, who can hold a copy and how recovery works. A flat claim with no mechanism, or a password-reset email that restores your files, is a reason to be doubtful.

Sources

VELDRIN

Run Veldrin through the key-origin test before you trust it

See Veldrin on the App Store