Re: [w3ctag/design-reviews] Broadening the user base of WebAuthn (Issue #686)

Sorry, there's something wrong with my GitHub notification emails and I missed a prior update:

> My thinking was that the user could have a device-bound key which is used to sign the public key of an ephemeral key pair that gets synced. When registering with a site, the device-bound key would be required to be present (to sign a site-specific ephemeral key that's generated on the fly), and the site would learn the public key of the device-bound key.
>
> However during normal usage, the site would accept a signed nonce from a key that is itself signed by the device-bound key, so the device-bound key need not be present. The model here is how CA's use an intermediate key to sign certificate requests, keeping their root key securely offline (or DNSSEC using zone-signing keys and key-signing keys).

I think the difference here is that the device-bound key wouldn't be involved in "normal usage". We believe that, for the majority of sites, a synced WebAuthn credential has good advantages over a password. But we hear from some sites that they really want a device-bound key as an additional signal. However, if the device-bound key simply signed a synced key then, at sign-in time, there's no difference between a known device signing in, an a leaked synced key being used with a replayed signature by a device-bound key.

In order to prevent replays, you would want the signature by the device-bound key to also sign over a nonce—proving that it was exercised for each sign-in request. At which point you have a something very similar to the current proposal.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/w3ctag/design-reviews/issues/686#issuecomment-1054849530

You are receiving this because you are subscribed to this thread.

Message ID: <w3ctag/design-reviews/issues/686/1054849530@github.com>