W3C

SPC Task Force

31 May 2021

Attendees

Present
Anne Pouillard (Worldline), Ian Jacobs (W3C), Jean-Carlo Emer (Stripe), Rolf Lindemann (Nok Nok Labs)
Regrets
Most people due to US holiday
Chair
Ian
Scribe
Ian

Meeting minutes

https://github.com/w3c/secure-payment-confirmation/blob/gh-pages/scope.md

https://github.com/w3c/secure-payment-confirmation/blob/gh-pages/scope.md#user-stories

Enrollment of multiple instruments with one authentication

<Rolf> Do we already cover the ability to use the underlying credential for traditional authentication?

https://github.com/w3c/secure-payment-confirmation/blob/gh-pages/requirements.md#fido-considerations

FIDO credentials should be "enhanceable" to SPC Credentials.

SPC credentials should also usable as ordinary FIDO credentials. See issue 39.

In-transaction enrollment, later authentication same merchant

IJ: can you optimize the UX to enroll+authenticate/sign

Rolf: Technically what you expect is after registration of credential. ID&V steps often prone to fishing. So registration time has a different security characteristics
… session binding assurance level not really known at this point
… how do you know it's the "same session". Cookie? stronger than that?
… binding assurance level might be relevant for this one

<Rolf> https://www.digital.govt.nz/standards-and-guidance/identification-management/identification-management-standards/binding-assurance-standard/

Authentication with out-of-band authenticator

Rolf: Best thing would be to be able to send transaction details to the out-of-band authenticator
… Level 1 extension for transaction confirmation technically is this kind of approach
… the concept is the same even if not same data as SPC
… it doesn't matter how you send the data, but would be great if you could do so
… what we do today is that the browser displays the transaction text
… the browser has a privileged position (since it can talk to the authenticator), this is the security that we are leveraging
… if you send the transaction text to the authenticator it would be even higher security level
… the browser should be able to *understand* whether the text will be displayed by the authenticator
… the browser needs to be able to say "Check your smartphone" in the dialog...that at least needs to be discussed
… I can send push notification that launches an app that talks to a server to trigger SPC
… you can trigger it through out-of-band mechanisms like QR codes

Jean-Carlo: In this case, would be still know user verification was formed?

Rolf: Yes, you get an assertion from the authenticator.
… suppose I'm on desktop without a registered platform authenticator
… it might know that I have a smartphone that supports SPC
… so it could send push notification to app or OS or browser on my smartphone to trigger spc on my device
… the browser on my phone would ask for confirmation which would generate the signed assertion
… we might not need to standardize the out-of-band behavior (since different on each platform)

IJ: What are security requirements for shipping to the phone?

Rolf: Merchant has the data (currency, credential ID)
… if SPC returns null, the merchant's JS can tell the merchant server.
… ideally, if my PC browser is synced to my phone, then the BROWSER might know I have an SPC credential on my Android device. In which case the display can be shipped to my phone where I can sign it.
… that would be a browser vendor decision
… multiple channels to ship the display info to another authenticator: CABLE, known authenticator since same vendor
… CABLE-connected is still considered "Roaming"

<Rolf> caBLE - cloud assisted BLE

next meeting

7 June

No meeting 14 June

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).