Popular Password Managers May Access Your Vault: Research Findings

Password managers are widely promoted as “zero-knowledge” solutions.

Companies claim that even if their servers are compromised, vault contents remain unreadable — even to the providers themselves.

However, recent research suggests that under certain conditions servers may gain access to encrypted user data.

When Zero-Knowledge Falls Short

Researchers from ETH Zurich and USI Lugano examined several major password management services.

They discovered that vulnerabilities often appear when account recovery or sharing features are enabled.

If a server is controlled by an attacker, it can substitute cryptographic keys and potentially gain access to part or all of a vault.

One key risk involves account recovery mechanisms.

When a new member joins a family or organization, the client receives keys from the server to encrypt recovery data.

If those public keys are replaced, an attacker may decrypt the data and restore the account as the victim.

Similar risks exist during key rotation procedures, especially if key authenticity is not properly verified.

Shared records introduce additional complexity.

When access is granted, new encryption keys are generated.

If those keys are manipulated, the attacker may read or modify shared data.

To maintain compatibility with older clients, some services support weaker encryption schemes.

This may allow a downgrade attack, reducing overall protection.

Additionally, servers could reduce password hashing iterations, making brute-force attacks significantly easier.

Researchers do not recommend abandoning password managers.

However, they argue that the term “zero-knowledge” should not be treated as an absolute security guarantee.

Security ultimately depends not only on encryption strength, but also on implementation details and trust in infrastructure.