Passkeys solved the hard part. Because a passkey is bound to the site it was created for, there’s nothing to type into a convincing fake login page, which is what makes phishing so effective against passwords. That’s the FIDO2 and WebAuthn design working as intended, and it’s sound. What isn’t sound is nearly everything wrapped around it.
Where your login lives is now a platform decision
Create a passkey in iCloud Keychain and it follows you across Apple devices. Create one in Google Password Manager and it follows you across Android and Chrome. Windows Hello keeps its own. Third-party managers like 1Password and Bitwarden support passkeys too, but the prompt that pops up on your phone usually nudges you toward the built-in vault, and defaults win.
Passwords were miserable, but they were portable. You could copy one from anywhere to anywhere. Passkeys give that up in exchange for real security, and I’d make that trade. But only if leaving stays possible. The FIDO Alliance has been working on a standard for moving credentials between providers, and Apple has shown off import and export support in its recent OS releases. That’s the right direction. It’s also early, and in the meantime plenty of people are building vaults they can’t easily carry out.

Recovery is the real test
Nobody judges a login system on a good day. They judge it the day the phone goes in a lake.
Passwords had a recovery story everyone understood: click “forgot password,” check your email. For passkeys, recovery usually means whatever your platform account can do, which turns your Apple or Google account into a single point of failure for everything else you own online.
The fallback door is still open
Here’s what bothers me most. Many sites now offer a passkey but leave the password and the SMS reset flow switched on behind it. That’s understandable, since support teams don’t want a flood of locked-out users. But an attacker doesn’t care how strong your front door is if the side door still takes a password. Until sites can retire the old method, the phishing resistance is partly theoretical.
This is a design problem more than a technical one, and it’s the kind the industry tends to solve slowly. The passkey itself is fine. I’m less sure what happens to the person who sets up two hundred of them, loses access to the one account that holds them all, and finds out the recovery path was never really built for them.