On Thu, 23 May 2024 at 20:58, Ilari Liusvaara <[email protected]>
wrote:

> On Thu, May 23, 2024 at 07:54:42AM -0500, Orie Steele wrote:
> >
> > As an author I support adoption.
> >
> > The main motivation I have for working on HPKE, is to ensure that
> encrypted
> > JWT and CWT use cases that want to use PQ algorithms have a consistent
> path
> > to achieve this.
> >
> > Both JWT and CWT share alignment regarding confirmation methods, so it is
> > important that they share algorithms as well, and with a few exceptions
> > that's mostly true today.
>
> Adding native PQ support is actually even easier than adding HPKE.
>

Adding a PQ/T Hybrid KEM in Hybrid Public-Key Encryption (HPKE) is much
easier, as detailed in
https://www.ietf.org/archive/id/draft-reddy-cose-jose-pqc-hybrid-hpke-00.html
.


>
> > I've implemented the current drafts for JOSE and COSE, I think it's been
> > beneficial to both specifications to compare the approaches.
>
> I have implemented what I think is still the current COSE draft. Found
> it easy. I have not even tried to implemented the JOSE draft because of
> a few issues I find rather nasty.
>

We have already addressed all your comments. Please review the latest
version of the draft to verify if there are any remaining comments.


>
>
> > It's true that currently DHKems feel very similar to ECDH-ES, but PQ or
> > hybrid KEMS won't.
>
> All KEMs feel very similar to ECDH-ES.
>
> The three operations ECDH-ES performs are _exactly_ the KEM keygen(),
> encaps() and decaps() operations!
>
>
> Now, it is not possible to just use ECDH-ES due to some technical
> details. But one can just easily clone the algorithms with just the
> technical details suitably changed.
>
> And then register the key subtypes for PQ keys to get complete usable
> PQ support.
>

The challenges with using hybrid schemes is different from using PQ KEM
algorithms alone.


>
>
> > Having a framework (JWT/CWT) in place, where JOSE and COSE can share PQ
> > algorithms will reduce complexity in the long run, and enable easier
> > security analysis and migration.
>
> Native PQ support can trivially share the algorithms.
>
>
> > There was also a presentation on designated verifier signatures at 119,
> > where HPKE was discussed in that context as well. I believe that direct
> > mode auth HPKE JWEs might help enable post quantum KEMs to be used for
> > those same use cases, whereas building on ECDH-ES + MAC as was recently
> > proposed would need more changes to support post quantum or hybrid
> > algorithms... So it's possible that this might save the need for future
> > algorithm registrations which might be requested to support redudiable
> > digital credentials use cases.
>
> Sorry, HPKE Auth mode is not supported for post-quantum.
>
> ... Because nobody knows how to create safe and usable authenticated
> post-quantum KEM!
>

To achieve an Authenticated Key Exchange, two full PQ KEM exchanges are
required, and their results must be combined to form a single shared
secret. This process is not feasible with JOSE/COSE.

-Tiru


>
>
>
>
> -Ilari
>
> _______________________________________________
> jose mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>
_______________________________________________
jose mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to