W3C

Web Payments Working Group

05 December 2024

Attendees

Present
David Benoit, Bjorn Hjelm (Yubico), Doug Fisher (Visa), Gerhard Oosthuizen (Entersekt), Grégoire Leleux (Worldline), Gustavo Kok (Netflix), Henna Kapur (Visa), Ian Jacobs (W3C), Jean-Luc di Manno (FIME), Jorge Vargas (Discover), Nakjo Shishkov (Netcetera), Praveena Subrahmanyam (Airbnb), Rogerio Matsui (Rakuten), Rouslan Solomakhin (Google), Sami Tikkala (Visa), Sidd Gothoskar (PayPal), Sharanya Chandrasekaran (PayPal), Slobodan Pejic (Google), Steve Cole (MAG), Sue Koomen (American Express), Vasilii Trofimchuk (Block)
Regrets
NickTR, Stephen McGruer
Chair
Ian
Scribe
Ian

Meeting minutes

Browser-based key in SPC next steps

(Slobodan Pejic walks through proposed changes to SPC for a BBK)

Slobodan: We plan to augment the payment extension in webauthn to communicate the BBK

Slobodan: Some development work is happening in Canary that has not quite landed but should be available soon in Canary (behind a flag)

Ian: Any surprises as you began to work on this?

Slobodan: Initially we thought we would have multiple signature, but will only have one
… we had thought about algorithm transitions and thought about cross-signing but that's rare.

Doug: Thanks for the update....I didn't quite understand the bullet point on registration.

Slobodan: When we generate a passkey in a 1p context, it's a different context from the assertion. The browser-bound signature should be available when the passkey is created.
… info would be in the client extension outputs

Ian: Anything changed in the Chrome built-in UX?

Slobodan: I don't have details there.

Ian: Are keys always created?

Slobodan: Yes, if payment extension is set. And may depend on availability of platform support.

Ian: Any updates on attestation?

Henna: We wanted to go back to check on threat model. The general concern is that when the device-bound key is created, is there a way for it to come directly from the platform (rather than the browser)?

Slobodan: Attestation in any case will not be part of first build

JeanLuc: How is the lifecycle of the BBK managed?

(See our requirements doc)

Slobodan: We have some options on this. E.g., we might delete and recreate a key if signing algo changes
… and treat it as a new key

Jean-Luc: If you create a new BBK for a credential, is there a way to keep a trace of the previous one for the purpose of chaining?

Rouslan: That won't work, for example, if the user has multiple devices.
… regarding user ability to delete the BBK, the user can do things like "clear browser data' and that should clear the keys
… switching profiles should also work

Ian: Slobodan, do you have any questions for the group that would be helpful for you?

Rouslan: General question is whether doing this will satisfy device binding requirements for the industry.
… we're trying to provide an approach to bind a key to at most one device.
… what in the industry would need to chance to accommodate this new key (e.g., in 3DS)

Gerhard: I think there's a mandate that something like dynamic linking must be done. If the device bound key is used to sign a payload I think the requirement will be met

Doug: From a 3DS side we'll be reviewing this and confirming this; I think it's not likely necessary since this is being done as an extension.

Doug: I assume the key is being created by the browser. Could it be created in the browser in a way that is protected from the JS layer?

Rouslan: Yes, it's protected from the JS layer.

Gerhard: Is device bound key merchant bound?

Ian: The BBK reqs doc currently says: "In private browsing mode, if a user can access passkeys when using SPC, the user agent should also return the BBK."

Slobodan: That still resonates for me.

Gerhard: +1 to being able to use in incognito mode.

Slobodan: Regarding cross-origin access, the key will be available cross-origin.

Gerhard: How complex would it be to add this concept to a payment handler, to complement other payment mechanisms.
… I'm talking with industry players who could be interested in a BBK for payment handlers themselves.

Ian: Different context than SPC, so we should separate that discussion

Jean-Luc: How would the signal about the device binding mechanism would be reflected?

Ian: Would you expect this signal about how the key was managed to be available in the first release of support?

Slobodan: That is not currently in the proposal because it may overlap with the attestation discussion.

Jean-Luc: For financial information, this information is very important.

Slobodan: I can take that as feedback. So far the initial proposal does not include that.

Henna: Regarding usefulness to industry, I think yes. If we can prove that the key is coming from a TPM and bound to a device, it should be ok.

Jean-Luc: I see synergies with DBSC. Is this key isolated from other initiatives?

IJ: From TPAC discussions I took away two aspects to that question: (1) The keys themselves should not be shared between the two APIs because there are different expectations about them in the two APIs (e.g., one says keys should be available cross-origin, the other says not to do that.). (2) The technologies should all be considered as part of building a good experience, because having multiple keys available may lead to better UX (e.g., having established a previous DBSC session on an origin might mean no step-up is required the first time the RP sees a new BBK).

Doug: When might this be available for testing?

Slobodan: I estimate "early Q1" in Canary behind a flag.

Payment request question about camera access in handler

w3c/payment-handler#420

Ian: Limitations today may mean you can't do facial recognition in a payment handler. That's the issue that was raised.

rouslan: We've not received feature requests from primary partners.
… are there partners who want to use payment handlers who want to experiment with this, please let us know.

Ian: Gerhard, is that interesting to the people you're talking to?

Gerhard: Absolutely. Selfie identification in payment handler would be an important capability.
… would it be a disqualifying thing if we can't? Probably not. But it would help.
… the reason I'm looking into payment handlers relates to FIDO adoption in the banking industry.
… so being able to use a camera for other types of authentication is important.

Next meeting

Gerhard: I will be presenting on payment fraud at the Antifraud CG on Friday at noon ET

Next WPWG call: 16 January 2025

Minutes manually created (not a transcript), formatted by scribe.perl version 238 (Fri Oct 18 20:51:13 2024 UTC).