Academic research demonstrates attacks against major cloud password managers
Vulnerability Disclosure · 2 sources · Updated 5mo ago · First seen Feb 23, 2026

Researchers from ETH Zurich and the Università della Svizzera italiana, led by Prof. Kenneth Paterson, published findings demonstrating 27 successful attacks against major password managers Bitwarden, LastPass, and Dashlane under a malicious server model, where an attacker has compromised the provider’s server. The work challenges the practical guarantees implied by “zero-knowledge encryption,” showing that if the server can tamper with what the client receives, some clients may fail to adequately verify integrity and binding between encrypted vault data and associated metadata, enabling vault contents to be exposed or misdirected.
The reported techniques include issues described as missing ciphertext integrity and insufficient cryptographic binding of fields (e.g., URL metadata not being tightly bound to the encrypted secret), enabling attacks such as field-swap scenarios where a decrypted password could be sent to an attacker-controlled domain during normal client behavior (e.g., fetching a site icon). Additional attack paths discussed target password-manager features beyond basic storage—such as account recovery, sharing, and auto-enrolment into organizations—reinforcing that password-manager security depends not only on encryption at rest but also on robust client-side validation and threat models that account for server compromise; broader commentary also notes recent, compounding weaknesses in the password ecosystem, including password-manager design assumptions and other emerging password-related risks.


