New Pass-ta-key attack reveals all the things we didn't know about passkeys
A clever exploit against Google Password Manager on Windows exposes the fine print behind passwordless security's biggest promise.
Written by OutOfToken AI
August 11, 2026 · 4 min read · Synthesized from reporting by Ars Technica · How this works
Passkeys were supposed to be the thing that finally killed the password: unphishable, unshareable, un-stealable. A new attack called Pass-ta-key just found a crack in that promise, at least on one platform. It doesn't break passkeys themselves — it exploits the software wrapped around them.
The pasta-themed exploit
Discovered by Arie Olshtein, a researcher at Palo Alto Networks' Unit 42, Pass-ta-key is a name that mashes up 'passkey,' 'pass the key,' and a wink at a plate of pasta. Playful branding aside, the research is serious: it shows how malware running on a Windows machine can extract passkeys stored in Google Password Manager (GPM) for Windows. Once extracted, those passkeys can be replayed to hijack the accounts they were meant to protect.
Why this undercuts the whole pitch
Passkeys are built on public-key cryptography specifically so the private key never leaves your device and can't be phished, guessed, or copied remotely. Google's own documentation leans hard on that pitch, describing passkeys as inherently resistant to sharing or theft. Pass-ta-key doesn't disprove that cryptography — it targets weaknesses in how GPM handles storage and synchronization on an already-compromised Windows machine.
"The attack only works if malware is already running on your device — which is precisely the scenario passkeys were designed to make irrelevant."
The nothingburger caveat
That's the catch security researchers and outlets covering Pass-ta-key keep circling back to. If an attacker already has malware executing on your Windows machine with the access needed to read GPM's data, they likely have easier paths to your accounts than reverse-engineering a passkey sync mechanism. The bigger headline isn't 'passkeys are broken' — it's that the software layer syncing them across devices carries its own attack surface, separate from the cryptography underneath.
Why Windows gets treated differently
The research zeroes in on Windows because that's where GPM's passkey handling reportedly diverges from how Google and other vendors manage passkeys on macOS, ChromeOS, or mobile. Passkey implementations aren't governed by a single uniform standard for storage and sync across every OS; each platform's password manager and OS-level credential store makes its own choices about encryption, caching, and how synced credentials get materialized locally. That inconsistency is exactly what let Unit 42's researcher find plaintext data readable enough to manipulate the cloud authenticator on Windows specifically.
Pass-ta-key won't send anyone rushing back to passwords, but it's a reminder that passwordless security isn't a single monolithic guarantee — it's a stack of implementations, and some layers are sturdier than others. Expect scrutiny of how password managers sync and store passkeys across platforms to intensify, especially on Windows, where this weakness surfaced first. The cryptography holds; the surrounding software still has homework to do.
Editorial Note
The research corroborates the core technical claims: Pass-ta-key's discovery by Arie Olshtein at Unit 42, the attack's mechanics on Windows GPM, and the fundamental security properties of passkey cryptography. Sources consistently support the article's nuanced position that the attack exploits software-layer weaknesses rather than breaking the underlying cryptography, and that the threat requires pre-existing device compromise. However, the specific claim about Windows-specific implementation divergence from other platforms lacks substantiation in the provided sources.
Claim Tracker
AI-assessed
Confirmed by Source 1 (Ars Technica), Source 2 (BleepingComputer), and Source 3 (Palo Alto Networks Unit 42)
Confirmed by Source 1 (Ars Technica): 'how Pass-ta-key could obtain all passkeys stored in the Google Password Manager app (GPM) for Windows when it's running on a machine infected with malware'
Supported by Source 2 (BleepingComputer): 'they cannot be guessed, reused, or easily stolen through phishing' and Source 5 (Malwarebytes) on the cryptographic basis
Confirmed by Source 4 (PCMag): 'if the device the passkey is stored on is compromised, that can be a real problem' and Source 5 (Malwarebytes) acknowledging the software weakness caveat
The article claims Windows handling 'diverges' from other platforms, but the provided sources do not contain comparative details about implementation differences across operating systems
Ask AI about this story
// discussion
sign in to join the discussion