39 Ways to Bypass Passkeys Without Breaking FIDO2 Cryptography

Passkeys were conceived as a replacement for passwords: public-key cryptography, binding to the legitimate service, a private key that never leaves the device. This really does shut down many classic phishing and credential-theft scenarios. But, as a publication (sponsored by Token) points out, the conversation about security quickly shifted onto different ground.

As of now, at least 39 methods, techniques and attack scenarios against passkeys and the infrastructure surrounding them have been publicly described. Many already have working proof-of-concept code, and some of the tricks show up in real-world attacks. The researchers' key takeaway: FIDO2 cryptography can remain entirely intact while the account it protects is already compromised.

The target isn't the key itself, but everything around it

The authentication ceremony crosses many trust boundaries: the web application, the browser, the operating system, the password manager, cloud sync, the mobile device, the Bluetooth channel, the recovery and registration processes, the support desk and, finally, the human being who confirms the login. Almost every one of these layers gets attacked.

SpecterOps research titled Pass the Passkey shows that a malicious application doesn't need to extract the private key at all. It can ask the legitimate WebAuthn infrastructure to generate a signed assertion: the user sees the familiar Windows authentication dialog, passes the check—and the attacker gets a ready-made result. The key never left secure storage, nobody broke the cryptography, yet the authentication process was successfully hijacked.

The login prompt is also an attack surface

Some techniques target the interface itself: bombarding the user with requests, spoofing application metadata, overlaying fake windows, phishing over remote desktop. This repeats the already familiar problem of MFA push notifications: once confirming a login turns into a routine gesture, it can be fabricated, replayed or timed just right. Phishing resistance at the protocol level does not mean resistance to deception at the level of the OS, the browser and the interface.

Sync and recovery

The attack surface expands sharply as soon as a passkey can be synced, exported, recovered or moved between devices: compromise of cloud storage, takeover of an Apple or Google account, theft of data exported from KeePassXC or Bitwarden, malicious browser extensions. This is essentially an architectural problem, not a cryptographic one.

A separate category consists of attacks that don't steal an existing key at all, but create a new one: “shadow” passkeys, vishing during registration, support-desk takeover, SIM-based recovery. Having gained enough control over an account, the attacker simply registers their own key on their own device, and the legitimate service hands them valid credentials all on its own.

Conclusion

The very existence of 39 methods, in the authors' view, actually confirms the strength of FIDO2: attackers go around the cryptography precisely because breaking it directly is too expensive. As a solution, the article predictably proposes dedicated biometric hardware with no screen, no browser and no cloud sync—but the general lesson is universal: phishing-resistant authentication is useless if the registration, replacement and recovery of keys are protected more weakly than the login itself.

Source: Bleeping Computer