Pass-ta-key Attacks Expose Google Chrome and Password Manager on Windows to Passkey Theft and MFA Bypass

Pass-ta-key Attacks Expose Google Chrome and Password Manager on Windows to Passkey Theft and MFA Bypass

Executive Summary

Recent research by Palo Alto Networks Unit42 and corroborated by RSA Security has revealed a new class of attacks—collectively dubbed "Pass-ta-key"—that target the security of synced passkeys, particularly in Google Password Manager and Google Chrome on Windows. These attacks allow malware on a compromised endpoint to recover synced private keys or bypass phishing-resistant multi-factor authentication (MFA), undermining the core security promises of passkey-based authentication. While no mass exploitation has been observed in the wild, the attack techniques have been demonstrated in controlled environments and highlight critical architectural weaknesses in current passkey synchronization implementations.

Technical Information

The "Pass-ta-key" attack family exploits the way Google Chrome and Google Password Manager handle passkey synchronization on Windows systems equipped with a Trusted Platform Module (TPM). The attacks do not require privilege escalation and can be executed by malware running as a standard user. There are three primary variants:

The first, known as the "Pass-ta-key Attack," leverages malware to extract the device’s identity key from the local Chrome profile directory (%LocalAppData%\Google\Chrome\User Data\<Profile>\Sync Data\LevelDB). The key, wrapped_identity_private_key, is accessible to user-level processes and can be used to sign authentication requests via Windows CNG APIs. This enables the attacker to impersonate the victim’s device to Google’s Cloud Authenticator and authenticate as the victim without any user interaction or device unlock.

The second variant, the "Silver Pass-ta-key Attack," involves malware deleting or invalidating the victim’s passkey_enclave_state file. This forces Chrome into a uv_key_pending state during device re-registration. The attacker can then register their own user verification (UV) key, which is accepted by Google’s Cloud Authenticator without attestation. This allows the attacker to authenticate from any device, not just the compromised one, with no further access to the victim’s device required.

The most severe, the "Golden Pass-ta-key Attack," targets the Security Domain Secret (SDS)—the master key used to encrypt all synced passkeys. During device registration or recovery, Chrome retrieves the SDS from Google’s Trusted Vault and stores it in process memory. Malware can dump this memory and extract the SDS, enabling the attacker to decrypt all current and future passkeys for the account. This results in persistent and total compromise, as the attacker can sign authentication challenges for any relying party.

All three attack variants exploit architectural flaws in the passkey synchronization and onboarding flows, rather than a single code bug. The attacks are specific to Google Chrome on Windows with TPM and do not currently affect other platforms or browsers.

Exploitation in the Wild

To date, there have been no public reports of mass exploitation of these vulnerabilities. However, the attack techniques have been demonstrated in controlled environments by Unit42 researchers. All attacks require malware to be present on the victim’s device, but do not require privilege escalation. Google has responded by removing the SDS from Chrome logs, but the key remains in process memory, leaving the attack surface partially exposed. Some relying parties, such as eBay, have patched their systems to properly validate the UV flag after responsible disclosure.

APT Groups using this vulnerability

No specific Advanced Persistent Threat (APT) group attribution has been made as of this report. However, the techniques align with those used by credential theft and espionage-focused actors. The attacks are within the capabilities of groups known for targeting authentication mechanisms, such as APT29 and APT41. Organizations in regulated industries and those with high-value targets should be especially vigilant.

Affected Product Versions

The affected products are Google Password Manager in Google Chrome on Windows. All versions of Google Chrome for Windows supporting passkey sync and Google Password Manager, on systems equipped with a TPM, up to at least the latest stable release as of August 2026, are vulnerable. No specific Chrome version numbers or build numbers have been excluded or patched as of this report. The attacks are architectural and affect the design of passkey sync and onboarding, not a single code bug. Other platforms, such as macOS, Linux, Android, and iOS, are not affected according to current public research. Similarly, other browsers like Edge, Firefox, and Safari are not known to be affected.

Workaround and Mitigation

For relying parties, it is critical to enforce userVerification = required and validate the UV flag in all authentication responses. Attestation of newly registered device keys should be validated to prevent unauthorized key registration. Vendors should prevent exposure of master keys (SDS) in client memory or logs and harden onboarding and recovery flows. Enterprises are advised to prefer device-bound or hardware-backed passkeys over synced passkeys for workforce and privileged accounts, monitor for abnormal passkey usage and device re-registration events, and deploy endpoint protection to detect malware targeting browser credential stores. Regularly review authentication logs for abnormal patterns, such as authentication from new devices or locations without corresponding user activity.

Indicators of Compromise

Indicators of compromise are point-in-time and should be validated before enforcement. No public indicators of compromise were available at the time of writing.

References

Palo Alto Networks Unit42: Pass the Passkey, RSA Security: Pass-ta-key Attacks, The Hacker News: Google Password Manager Attacks, Reddit: Multiple Flaws in Google’s Synced Passkey Implementation, NVD

Rescana is here for you

Rescana provides a comprehensive Third-Party Risk Management (TPRM) platform that empowers organizations to continuously monitor, assess, and mitigate cybersecurity risks across their vendor ecosystem. Our platform delivers actionable intelligence and automated workflows to help you stay ahead of emerging threats. We are happy to answer questions at info@rescana.com.