Skimming papers like this I always try to map findings to a user threat model.
If the attacks are mostly “malicious/compromised client + browser extension” or “malicious local process”, then the uncomfortable takeaway is that any vault in the same trust zone is in trouble (the manager can’t defend secrets from the machine that’s currently decrypting them).
If there are server-side / sync-layer attacks that let a provider or MitM cause unsafe behavior (rollback, confused deputy, malicious item injection, etc.), then that’s where vendors can materially improve: stronger integrity checks, monotonic versioning, explicit trust prompts, safer parsing, independent verification.
Would love a quick taxonomy summary from someone who’s read the whole PDF: which category dominates?
Skimming papers like this I always try to map findings to a user threat model.
If the attacks are mostly “malicious/compromised client + browser extension” or “malicious local process”, then the uncomfortable takeaway is that any vault in the same trust zone is in trouble (the manager can’t defend secrets from the machine that’s currently decrypting them).
If there are server-side / sync-layer attacks that let a provider or MitM cause unsafe behavior (rollback, confused deputy, malicious item injection, etc.), then that’s where vendors can materially improve: stronger integrity checks, monotonic versioning, explicit trust prompts, safer parsing, independent verification.
Would love a quick taxonomy summary from someone who’s read the whole PDF: which category dominates?
“We present 12 distinct attacks against Bitwarden, 7 against LastPass and 6 against Dashlane.” oof