Encryption & Key Administration
,
Identification & Entry Administration
,
Safety Operations
‘Malicious Server Risk Mannequin’ Threatens ‘Zero Data Encryption’ Ensures

Claims by main stand-alone password managers that their implementation of “zero data encryption” means saved passwords can face up to the worst of hacker assaults are vastly overblown, say educational safety researchers.
See Additionally: Cracking the Code: Securing Machine Identities
A crew of 4 hailing from Switzerland’s ETH Zurich and the USI Università della Svizzera italiana discovered that safety ensures by cloud-based password supervisor software program provided by Bitwarden, Dashlane and LastPass aren’t as marketed.
The three password managers – chosen as consultant samples, based mostly on market share, their skill to entry unobfuscated supply code for the merchandise, in addition to “the richness of the provided characteristic set and the range of approaches” – promote that password vaults ought to be secure even when hackers compromise the server that shops them.
All three of the distributors promise zero data encryption, which means they cannot see into the password vault. The researchers mentioned there is not any industry-accepted definition for what this implies: It is a advertising and marketing time period. Even so, “the promise is that even when somebody is ready to entry the server, this doesn’t pose a safety danger to prospects as a result of the info is encrypted and subsequently unreadable,” mentioned Matilda Backendal, an assistant professor at USI Università della Svizzera italiana who was a part of the analysis crew.
“We’ve now proven that this isn’t the case,” she mentioned.
Safety consultants have lengthy really helpful password managers, not least as a result of they can be utilized to generate and handle distinctive passwords for each completely different web site or service. The safety problem is that password managers might be stuffed with fascinating info.
“Because of the great amount of delicate knowledge they include, password managers are possible targets for skilled hackers who’re able to penetrating the servers and launching assaults from there,” mentioned Kenneth Paterson, professor of pc science at ETH Zurich, who co-authored the report.
All three password managers fell brief indirectly. “The assaults enable us to downgrade safety ensures, violate safety expectations and even totally compromise customers’ accounts,” the researchers wrote in a paper set to be offered in August on the thirty fifth annual USENIX Safety Symposium being held in Baltimore.
“Worryingly, nearly all of the assaults enable restoration of passwords – the very factor that the password managers are supposed to shield.”
Researchers subjected every password supervisor to a “malicious server risk mannequin,” by which they recognized 12 several types of assaults that succeeded towards Bitwarden, seven towards LastPass and 6 towards Dashlane.
Malicious servers can pose a danger resulting from how cloud-based password managers retailer and grant entry to a consumer’s encrypted password vault within the cloud.
When a consumer of such a service needs to retrieve a password, they authenticate to the service supplier, pull a duplicate of the vault onto their consumer and decrypt it utilizing their grasp password.
In comparison with a client-only password supervisor, cloud providers could provide a wide range of further options. These can embrace the flexibility to share passwords with household or coworkers inside a company, to entry passwords in a wide range of methods together with by an internet browser and cellular system, key restoration providers and for organizations to self-host their very own password servers.
The researchers mentioned that in lots of instances, the check server they created to imitate malicious habits not solely gave them entry to a consumer’s saved passwords, however allow them to change the saved password.
“We had been shocked by the severity of the safety vulnerabilities,” mentioned Paterson.
They grouped the issues they discovered into 4 classes: exploits of key escrow options that help single sign-in logins and account restoration, exploits of vault integrity, exploits of sharing options and exploiting backwards compatibility options.
The analysis crew notified all the distributors about their findings and shared proof-of-concept exploits, set a 90 day public disclosure deadline and mentioned they labored carefully to assist as they ready fixes. After notifying Bitwarden in January 2025, LastPass in June 2025 and Dashlane in August 2025, the researchers subsequently pushed again the disclosure deadline after a request from LastPass, which final July additionally awarded them two bug bounties and continued to maintain them carefully apprised of its efforts to patch the issues.
“For essentially the most half, the suppliers had been cooperative and appreciative, however not all had been as fast when it got here to fixing the safety vulnerabilities,” Paterson mentioned.
Info Safety Media Group approached Bitwarden, Dashlane and LastPass for remark. The researchers mentioned all the distributors deliberate to handle the vulnerabilities and their response publicly.
“Dashlane discovered no proof of exploitation associated to those points,” the corporate mentioned and pointed to a weblog publish it printed on Monday responding to the findings and detailing its repair.
“It is also essential to notice that the assaults recognized by the researchers require full compromise of a password supervisor’s servers, paired with a extremely subtle risk actor capable of execute cryptographic assaults, and for sure findings, both particular circumstances and/or a particularly important window of time,” it mentioned.
The researchers additionally carried out an preliminary evaluation of 1Password, and mentioned they discovered – and reported – methods its software program may very well be exploited utilizing a malicious server. In addition they famous that the software program does require a “secret key” to entry a vault – not only a grasp password – which ought to safeguard it towards brute-force assault dangers discovered within the different password managers. 1Password mentioned it is reviewed the analysis and located it does not element any assaults not already documented within the firm’s personal safety design white paper, which notes: “At current, there is not any sturdy methodology for a consumer to confirm the general public key they’re encrypting knowledge to belongs to their supposed recipient. As a consequence, it will be potential for a malicious or compromised 1Password server to offer dishonest public keys to the consumer and run a profitable assault.”
“We’re dedicated to repeatedly strengthening our safety structure and evaluating it towards superior risk fashions, together with malicious-server eventualities like these described within the analysis, and evolving it over time to take care of the protections our customers depend on,” mentioned Jacob DePriest, 1Password CISO and CIO.
The researchers mentioned their findings can possible be used to focus on different cloud-based password managers, and well-resourced nation-states could already be doing so. In addition they mentioned that an off-the-cuff group of password administration distributors has been conscious of their analysis and findings since January 2025, which means distributors of instruments they did not overview can even have had the possibility to search out and tackle any such flaws in their very own choices.
“We can not exclude the chance that our assaults had been already identified to superior risk actors – in any case, now we have realized from the Snowden revelations that nationwide safety businesses are routinely tasked with penetrating techniques like those we analyze and are keen to conduct energetic assaults on targets,” the paper says.
The researchers mentioned they hope to boost the safety baseline being provided by cloud-based password managers, by “pushing their distributors to both enhance safety or make clearer statements about what safety their techniques truly present, in order that prospects can choose (maybe with the assistance of professional guides) whether or not the merchandise meet their necessities or not.”
“My suggestion remains to be to make use of a password supervisor, and I do not assume customers ought to be afraid of cloud-based ones,” Backendal mentioned.









