"Military-grade encryption."
That phrase appears on the homepage of almost every password manager. It sounds impressive. It means almost nothing unless you understand what it refers to and, more importantly, what other security properties need to be in place for that encryption to actually protect you.
Encryption is one layer. A password manager needs at least three layers working together to be genuinely secure: strong encryption, a zero-knowledge architecture, and end-to-end encryption. Let's look at what each one does and why removing any one of them undermines the others.
AES-256: The Encryption Standard
AES stands for Advanced Encryption Standard. The "256" refers to the key length — 256 bits. AES-256 is the same encryption standard approved by the U.S. National Security Agency for handling Top Secret data. It's used by banks, militaries, and governments worldwide.
Here's what AES-256 does in a password manager: it takes your vault — all your passwords, notes, credit card numbers, everything — and scrambles it into ciphertext that is mathematically infeasible to decrypt without the key. With current computing technology, brute-forcing a 256-bit key would take longer than the age of the universe. Even quantum computing, the most commonly cited future threat to encryption, doesn't meaningfully reduce the security of AES-256.
But AES-256 is only as strong as the key that unlocks it. And that's where the next layer comes in.
The Key Derivation Problem
Your master password generates the encryption key. This process is called key derivation, and the function used matters enormously.
Older password managers used PBKDF2 (Password-Based Key Derivation Function 2) with a relatively low iteration count — maybe 1,000 rounds. This was fine in 2010. It is not fine now. Modern GPUs can try billions of PBKDF2 iterations per second, meaning a weak master password can be cracked in hours if an attacker has your encrypted vault.
The current standard is Argon2id, which won the Password Hashing Competition in 2015. Argon2 is designed to be memory-hard, meaning it requires a significant amount of RAM for each attempt. This makes GPU-based attacks far less efficient because you can't run thousands of attempts in parallel on a single card.
When you evaluate a password manager, look for Argon2id with a memory cost of at least 64MB and a time cost of at least 3 iterations. If the tool uses PBKDF2, it should use at least 600,000 iterations. Anything less is a red flag.
This matters because of what happened in the LastPass breach. Attackers stole encrypted vaults and were able to brute-force weak master passwords offline because the key derivation was too fast. Strong encryption with weak key derivation is like a bank vault with a combination lock where the combination is your birthday.
Zero-Knowledge Architecture
This is the property that separates genuinely secure password managers from those that merely encrypt your data.
In a zero-knowledge architecture, the encryption key — derived from your master password — never leaves your device. The server never sees it. The vendor never sees it. When your vault syncs to the cloud, it's already encrypted with a key that only you possess.
This means the vendor mathematically cannot decrypt your data. If they're compelled by law enforcement, if their servers are breached, if a rogue employee wants to snoop — none of it matters, because the key isn't there.
Without zero-knowledge, encryption is theater. If the vendor holds the decryption key, then a breach of the vendor means a breach of your data, regardless of how strong the encryption is. The encryption just adds a step the attacker has to go through — a step the vendor's infrastructure may or may not protect.
You can verify zero-knowledge by checking whether the tool's security whitepaper or documentation explicitly states that the encryption key is derived locally and never transmitted. Open-source tools can be verified by code audit. Closed-source tools require you to trust the vendor's claims, which is why open-source password managers have an inherent advantage here.
End-to-End Encryption
End-to-end encryption (E2EE) is the property that ensures your data is encrypted from the moment it leaves your device until the moment it arrives on the receiving device — with no decryption in between.
In a password manager context, this matters for syncing. When you add a password on your laptop and it appears on your phone, E2EE means the data was encrypted on the laptop, transmitted as ciphertext, and decrypted on the phone. The server that relayed the data never saw the plaintext.
This sounds like it should be obvious. You'd be surprised how many "encrypted" services decrypt data on the server, process it, and re-encrypt it. That's encryption in transit, not end-to-end encryption. The server sees plaintext. If the server is compromised, your data is exposed.
E2EE and zero-knowledge are related but distinct. Zero-knowledge means the vendor can't decrypt your data. E2EE means the infrastructure can't decrypt your data in transit. You want both. One without the other leaves a gap.
How These Three Layers Work Together
Think of it as a chain. AES-256 encrypts your vault. Argon2id derives the key from your master password in a way that resists brute-force attacks. Zero-knowledge architecture ensures the key never leaves your device. End-to-end encryption ensures the ciphertext stays encrypted during sync.
Remove any link and the chain breaks:
Strong AES-256 with weak key derivation means a weak master password undermines everything. The encryption is perfect, but the key is guessable.
Strong encryption with zero-knowledge but no E2EE means your data is secure at rest but potentially exposed during sync if the relay server is compromised.
E2EE without zero-knowledge means the vendor holds a key that can decrypt your data, so a breach of the vendor still exposes you.
All three together means: your data is encrypted with a key that only you have, derived in a way that resists brute force, and protected end-to-end during transmission. That's the gold standard.
Verifying the Claims
Every password manager claims to be secure. Here's how to check:
Read the security whitepaper. Reputable tools publish detailed documentation of their encryption stack. If the whitepaper doesn't mention Argon2id, zero-knowledge, and E2EE by name, ask why.
Look for independent audits. Security firms like Cure53, Trail of Bits, and NCC Group regularly audit password managers. An audit doesn't guarantee security, but it means independent experts have reviewed the claims.
Check if the tool is open-source. Open-source doesn't automatically mean secure, but it means the claims can be verified. Closed-source means you're trusting marketing copy.
Try the NovelCrypt Password Vault, which uses AES-256 with Argon2id key derivation, a zero-knowledge architecture, and end-to-end encryption by default. Or compare it against other options in our password manager comparison.
The encryption is the easy part. Getting the architecture right around it is what separates a secure password manager from one that just claims to be.
