Good idea - work in progress
"So one of the things that we're seeing is the whole movement away from passwords to passkeys – a certificate-based authentication wrapped in a usability shrink wrap," Forrester VP and analyst Andras Cser told The Register.
- I think this is the best definition of passkeys I have seen. It does not try to force implementation limits like other definitions have. Best practices should be defining the implementations. But I have seen implementations which should not have been released yet, keep those implementations in beta until wrinkles have been ironed out. Just like modern software development though, public beta testing.
Gartner analyst James Hoover told The Register. "For device-bound FIDO2 keys, there is not currently a proven method of 'stealing' them, as the private key itself does not leave the device."
- WTF? Steal the device! Oh, I see Gartner, move on - pay no attention to the man in the ivory tower.
"With passkeys, we take that shared-secret model and just blow the whole model up, so there's nothing that can be shared," FIDO Alliance CEO and executive director Andrew Shikiar told The Register.
- Really now?! How does that work? Wait for it, wait for it ...
Then there are multi-device passkeys – synced credentials that allow users to log into apps on any of their devices and stored in a credential manager like Google Password Manager, iCloud Keychain, or open source Bitwarden.
- It looks to me like passkeys can be "shared".
These implementations are all over the map. I have passkeys in a FIDO2 USB/NFC hardware device, in a password manager, and in a Windows laptop. I could have in browsers, phone, phone browsers, etc. Some sites refuse to establish passkeys on some platforms be they Linux or password managers, non-Chromium browsers, or who knows. For the uninitiated, good luck understanding what a passkey is when all these ARE passkeys to the end user.
Windows requires a PIN to establish passkey use. The PIN is not for an individual passkey. It is for the Windows credential store. A password is not good enough when storing a passkey in the store - fair. But then it allows a four digit PIN - kinda weak. And it can be used in lieu of the password - Now were cooking with napalm! (hey kids: don't try that at home, toxic substances on the food, probably cook yourself too, and likely burn down the house) Oh, and I almost forgot, reset your PIN and passkey goes bye-bye. Good to see that MS has gone to great lengths to ensure their Windows passkey implementation is of equal quality to that of the OS. But I do think MS has contributed greatly to the "2 billion passkeys being used" between Windows and their other ?AAS offerings for varying interpretations of "used".
Other implementations have their issues as well. I cannot speak to Apple's, not having anything to do with that company since the Apple II line died out. Google, bitwarden, local banks, etc. all have chosen different ways to implement passkeys. Different passkey storage "devices" walk all over each other trying to be your favorite place to keep your passkeys. Some even allow set up of a catch-22^2 of passkey to access passkey to access passkey ad infinitum or so it seems since the chain of passkeys look so similar. Some sites offer passkey access or so they say, I have not been able to get one to allow any particular combination of software presented to establish a passkey, but they will text or email an access code instead of using a password, heck they will still demand you choose text or email delivery of a code even after providing the password. With security like this who needs locks? Wait 'til you see how these characters have implemented "lost authenticator recovery".
One final note. What you are is not a viable authentication factor, what you are IS your true identity. Other identifiers are usernames, ID numbers, names, where you have lived, where you went to school, what was your first car, mother's maiden name, things that make up you that do not change. In many instances Identity needs to be Authenticated and sometimes more assuredly than others. When the ID assurance level needs to be high, non-identity authentication factors should be used and should be changed if suspected to be compromised. Nothing is perfect, and implementation is key.