Hello,

> However there is still one major blocker for my use of it.  I still need an 
> algorithm for "HPKE-P256-SHA256-A256GCM" that used today, otherwise I cant 
> use the new standard when approved.  While I dont think the format should 
> determine if the HPKE cipher suite is valid (HPKE should do that), or if the 
> algorithm strength is sufficient for the intended use (this is 
> sender/receiver policy), we are already past that.  If we add another 
> "HPKE-X" algorithm or a way to handle adhoc combinations then it would work 
> for me.
> 

I still have the same problem.  I have a valid HPKE cipher suite, use JOSE for 
communication, but the proposed standard does not have a way for me to use it 
because there is not a HPKE_X algorithm available.  Can we add another entry 
for my existing algorithm?  Or should I pan on adopting the RFC, except for the 
missing algorithm names?

Regards,
-matt



> On Oct 11, 2025, at 7:04 AM, John Mattsson 
> <[email protected]> wrote:
> 
> Ilari Liusvaara wrote:
> 
> >The (pre-quantum) strengths of P-256 and AES-128 can not be directly
> compared because scaling is different (P-256 has quadric scaling and
> AES-128 has linear scaling). However, P-256 is clearly stronger of the
> two.
> 
> Yes, and the same is true for X25519, it is also (as far as we know) stronger 
> than AES-128 against classical attackers. Might be
> 
> >Even worse would be trying to to match strengths of confidentiality
> and authentication. See RFC9420 for an example of this going very
> wrong.
> 
> Agree. Typically, a better strategy is to go for simplicity and fewer options.
> 
> The worst and dangerous matching I have seen in practice is people justifying 
> using random nonces and a fixed key with AES-GCM just because the security 
> against distinguishing attacks is low (< 128). This leads to much lower 
> practical security. With a nonce collision an attacker can recover two full 
> messages and compromise the integrity for the rest of the sessions. With a 
> distinguishing attack, the attacker can recover a negligible fraction of a 
> single bit of plaintext….
> https://eprint.iacr.org/2024/1111
>  
> Cheers,
> John
> 
> 
> 
> matching the strengths against passive attacks on confidentiality and the 
> 
> 
>  
> From: Ilari Liusvaara <[email protected]>
> Date: Saturday, 11 October 2025 at 13:33
> To: [email protected] <[email protected]>
> Subject: [jose] Re: I-D Action: draft-ietf-jose-hpke-encrypt-12.txt
> 
> On Sat, Oct 11, 2025 at 12:31:57PM +0530, tirumal reddy wrote:
> > 
> > For the P-256, the equivalent symmetric security level is 128 bits, it
> > should be paired with the AES-128 algorithm for a matching security
> > strength.
> 
> The (pre-quantum) strengths of P-256 and AES-128 can not be directly
> compared because scaling is different (P-256 has quadric scaling and
> AES-128 has linear scaling). However, P-256 is clearly stronger of the
> two.
> 
> Even worse would be trying to to match strengths of confidentiality
> and authentication. See RFC9420 for an example of this going very
> wrong.
> 
> 
> 
> 
> -Ilari
> 
> _______________________________________________
> jose mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> _______________________________________________
> jose mailing list -- [email protected] <mailto:[email protected]>
> To unsubscribe send an email to [email protected] 
> <mailto:[email protected]>
_______________________________________________
jose mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to