Passkeys have been promoted as the safer replacement for passwords, with protection against phishing, credential reuse, and password leaks. Google has gone further than that, saying a passkey cannot be copied or accidentally handed to someone else. Security researchers have now shown three ways malware can work around those promises when passkeys are synced through Google Password Manager.
The three techniques are grouped under the name Pass-ta-key. They come from Unit 42, and BleepingComputer covered the disclosure. All three target Google Password Manager inside Chrome on Windows computers that have a Trusted Platform Module, and all three share the same prerequisite: malware has to be running on the victim's machine already.
The cryptography is not what broke
None of the attacks defeat the cryptography behind passkeys. The researchers went after the parts of the system that surround the key instead: device trust, account recovery, onboarding, and the way individual services check whether the user actually unlocked their device.
That distinction matters for how you read the rest of this. A passkey that cannot be phished is still a passkey that a compromised computer can be tricked into using.
Malware can pose as your trusted computer
The first technique uses Chrome's TPM-backed device identity to request a valid passkey response from Google's cloud authenticator. There is no administrator prompt, no fingerprint or PIN, no device unlock, and nothing the victim has to click. Google's service reads the request as coming from a computer it already trusts and hands back the authentication response needed to sign in.
Websites are meant to catch this. The response carries a flag confirming that the user verified their identity, and a site that requires verification should reject a response without it. Unit 42 tested that assumption and found the results uneven. GitHub rejected the attack. eBay accepted it, despite supposedly requiring verification, and closed the gap after the researchers reported it.
Silver Pass-ta-key plants the attacker's own verification key
The more advanced version of the attack can force Chrome to register a verification key that the attacker controls. Once that key is in place, it gets treated as proof that the victim entered a PIN or used biometrics. The payoff for an attacker is access from a completely different computer, which stays available after the victim's original device goes offline.
Golden Pass-ta-key steals the keys themselves
The last technique is the most serious one, because it targets the master secret used to encrypt every passkey synced through a Google account.
Researchers first found that secret sitting in plain text in Chrome's internal FIDO logs. Google removed it from the logs after disclosure. According to Unit 42 the secret still appears briefly in Chrome's process memory during device registration or recovery, and malware can pull it from there and decrypt the victim's synced passkeys.
A stolen master key could reportedly expose passkeys that already exist along with any created later. Unit 42 also notes that Google's current implementation offers no way to rotate or revoke that secret once it has been compromised.
What this does and does not change
Passkeys remain substantially safer than passwords against phishing and password leaks, and Google's documentation describes those advantages accurately. What the Pass-ta-key research adds is a narrower but real caveat. Malware that is already inside your computer does not have to break a passkey to abuse one. It can go after the infrastructure built around it.