I have made that update to the Complex Delegation bullet.

The update can be found here:
https://datatracker.ietf.org/doc/charter-ietf-oauth/

Apologies for the delay.

Deb Cooley
Sec AD

On Sun, May 10, 2026 at 10:45 PM Rifaat Shekh-Yusef <[email protected]>
wrote:

> From your description and from skimming your draft, the problem seems to
> be in scope of the OAuth WG even before the rechartering.
> Having said that, your document is not a specification; is your plan to
> submit a proposed specification to address all these objectives?
>
> Regardless, let's take this off this thread, which is focused on the
> charter.
>
> Regards,
>  Rifaat
>
>
>
> On Sun, May 10, 2026 at 2:48 PM Mohamad Khalil Yossif <[email protected]>
> wrote:
>
>>
>> Rifaat,
>>
>> Thanks - and I appreciate the proposed wording change. The shift from
>> "identity and authorization chaining" to "authorization of automated
>> agents working on behalf of users" is meaningful; it broadens the bullet
>> from inter-workload chaining to user-authorization for agent actions,
>> which is the harder problem.
>>
>> On RFC 9470: yes, I've read it, and it's the closest neighbor in the
>> OAuth family to what draft-yossif-psea-01 addresses - but I think they
>> sit on different axes. Let me try to make the distinction precise.
>>
>> RFC 9470 lets a resource server signal that the authentication event
>> behind the current access token is insufficient - too weak (acr_values)
>> or too old (max_age) - and triggers re-authentication that produces a
>> new token carrying updated acr and auth_time claims. The model assumes
>> an active session, the AS performs the authentication, and the resulting
>> evidence is encoded as claims inside the token.
>>
>> PSEA addresses a related but separate property: cryptographic proof,
>> generated at the moment of action, that a specific enrolled human
>> authorized that specific action with that specific payload - where
>> "specific enrolled human" means the biometric template registered for
>> this account at enrollment, not "someone who passed UV on this device".
>> The proof is bound to the action canonical form (amount, recipient,
>> dosage, etc.), produced inside the user's device hardware (Secure
>> Enclave / StrongBox), and verifiable independently of session state.
>>
>> The practical differences I see:
>>
>> 1. Binding axis. 9470 binds an authentication event to a token. PSEA
>>    binds a specific-human authorization event to a canonical action
>>    payload. A 9470-step-up token, once issued, can authorize N
>>    subsequent actions the user never saw; PSEA evidence is per-action
>>    and per-specific-human.
>>
>> 2. Session dependency. 9470 explicitly assumes an active session and
>>    an available AS at authentication time. PSEA is designed to operate
>>    post-session and, in some tiers, offline - with verification
>>    finalized server-side later - for environments where AS reachability
>>    cannot be assumed at action time (regulated workforce, healthcare
>>    dispensing, agentic execution under intermittent connectivity).
>>
>> 3. Audit persistence. PSEA evidence is constructed so that the proof
>>    of who authorized what survives independently of session and AS
>>    state - relevant to regulated regimes such as PSD2 SCA dynamic
>>    linking and eIDAS, which require per-transaction authorization
>>    evidence linked to amount and payee, retained for audit.
>>
>> 4. Trust root. In 9470, the AS is the authority that attests
>>    authentication occurred. In PSEA, the proof originates from the
>>    user's device hardware and is verified against device-trust state.
>>    The AS, if present, finalizes - it does not produce the
>>    human-authorization evidence.
>>
>> 5. Replay surface. A 9470 token is still a bearer credential within
>>    its acr context; a captured token can authorize unintended actions
>>    within scope. PSEA evidence is action-scoped by construction -
>>    capture of one proof yields no leverage on a different action.
>>
>> I am not claiming 9470 is deficient - for its problem (session-strength
>> signaling) it works well. The question is whether the rechartered
>> Complex Delegation bullet, in its new wording, is intended to also
>> accommodate per-action human-authorization evidence as a distinct
>> mechanism, or whether that should be pursued elsewhere (WIMSE, SPICE,
>> a future BoF).
>>
>> Happy to take the technical comparison off-list if that's more useful
>> than continuing on the charter thread.
>>
>> Best regards,
>>
>> [image: Mohamad Khalil Yossif, Founder & CEO of Yuthent]
>> Mohamad Khalil Yossif
>> Founder & CEO · Yuthent
>> PSEA Spec Author · IETF draft-yossif-psea
>> [image: Yuthent — Execution Authority Infrastructure]
>>
>> *T:* +972 50-931-1103 <+972509311103>
>> *E:* [email protected]
>> *W:* yuthent.com [image: LinkedIn]
>> <https://www.linkedin.com/in/mohamadkhalilyossif>
>>
>> [image: Yuthent — Cryptographic proof of human authorization at execution
>> time] <https://yuthent.com>
>> This email may contain confidential or sensitive information. If you are
>> not the intended recipient, any review, use, disclosure, distribution, or
>> copying is prohibited. If you received this message in error, please delete
>> it immediately.
>>
>> On 10/05/2026 20:55:57, Rifaat Shekh-Yusef <[email protected]>
>> wrote:
>>
>>
>> On Sun, May 10, 2026 at 1:23 PM Mohamad Khalil Yossif <
>> [email protected]> wrote:
>>
>>> Rifaat, Hannes, Deb,
>>>
>>> Thanks for the proposed text. The new framing — *"as automated agents
>>> increasingly act on behalf of users, organizations, or both"* —
>>> captures an important shift, and I appreciate that the Work Program
>>> acknowledges it explicitly under Complex Delegation.
>>>
>>> I'd like to ask a scoping question to understand whether a problem space
>>> I've been working on falls inside or outside the rechartered scope, before
>>> the telechat.
>>>
>>> The Complex Delegation bullet is described as *"new mechanisms or/and
>>> extensions for identity and authorization chaining to address scenarios
>>> where automated agents act across multiple administrative domains."* My
>>> reading is that this is principally about chaining the *identity and
>>> authorization of workloads* as a request traverses domains — i.e., who
>>> is acting and under what delegated grant — and that the Coordination
>>> section reinforces this by pointing service-to-service and multi-hop
>>> workload identity work toward WIMSE.
>>>
>> We are in the process of updating this specific bullet to the following:
>>
>>> Developing new mechanisms or/and extensions for authorization of
>>> automated agents working on behalf of users, including addressing scenarios
>>> where automated agents act across multiple administrative domains.
>>
>>
>>
>>> What is less clear to me is whether the charter intends to cover 
>>> *execution-time
>>> authority assurance* as a distinct concern: cryptographic evidence,
>>> produced at the moment of a sensitive action, that a specific human (not
>>> just a valid session or a properly delegated workload) authorized that
>>> specific action, with the action payload bound to the proof. This is
>>> adjacent to delegation but operates on a different axis — it is not about
>>> who holds the token or how authority is chained, but about whether the
>>> executing party can prove human authorization at action time, independent
>>> of prior session state.
>>>
>> I have not read your draft, but have you looked at RFC9470?
>> https://datatracker.ietf.org/doc/rfc9470/
>>
>> Is this somehow related to the ideas you describe in your draft?
>>
>> Regards,
>>  Rifaat
>>
>>
>> I ask because I authored draft-yossif-psea-01 (Post-Session Execution
>>> Assurance), which addresses exactly this gap, and I want to understand
>>> whether the natural home for follow-on work in this area is OAuth (under
>>> Complex Delegation or a future extension), WIMSE, SPICE, or somewhere else
>>> entirely. I am not asking the charter to call this out explicitly — I am
>>> asking the chairs and AD how they see the boundary.
>>>
>>> Concretely:
>>>
>>>    1. Does Complex Delegation, as currently scoped, contemplate
>>>    per-action human authorization proofs bound to the action payload, or is 
>>> it
>>>    limited to identity/authorization chaining between automated parties?
>>>    2. If the former is out of scope here, would the chairs see this as
>>>    WIMSE territory, SPICE territory, or material for a future BoF?
>>>
>>> Apologies if this has been discussed in a venue I missed. Happy to take
>>> the answer offline if it is more efficient.
>>>
>>> Best regards,
>>>
>>> [image: Mohamad Khalil Yossif]
>>> Mohamad Khalil Yossif
>>> CEO – Yuthent
>>> Founder – FitSnitch, Thesis, SwiftCrew, MK Digital, MK Electronics
>>> [image: Yuthent Logo]
>>>
>>> *T:* +972 50-931-1103 <+972509311103>
>>> *E:* [email protected]
>>> *W:* yuthent.com [image: LinkedIn]
>>> <https://www.linkedin.com/in/mohamad-khalil-yossif-4b9781163/>
>>>
>>> [image: Yuthent – Verify the Human] <https://yuthent.com>
>>> This email may contain confidential or sensitive information. If you are
>>> not the intended recipient, any review, use, disclosure, distribution, or
>>> copying is prohibited. If you received this message in error, please delete
>>> it immediately.
>>>
>>> On 10/05/2026 20:07:37, Rifaat Shekh-Yusef <[email protected]>
>>> wrote:
>>>
>>>
>>> ‪On Thu, May 7, 2026 at 1:21 PM ‫אלעזר פוקס‬‎ <[email protected]>
>>> wrote:‬
>>>
>>>> Hi Rifaat, Hannes, and Deb,
>>>>
>>>> Thanks for sharing the updated charter. I have a few concrete points
>>>> regarding boundaries and future scope that might be worth clarifying before
>>>> the IESG telechat:
>>>>
>>>> 1. Coordination with the SPICE WG: Given that the newly formed SPICE WG
>>>> is also tackling selective disclosure and digital credentials (SD-CWT), it
>>>> would be helpful if the OAuth charter explicitly defined the coordination
>>>> mechanisms. We should ensure there's no overlapping work, especially around
>>>> Metadata and Capability Discovery.
>>>>
>>>> I do not think that this needs to be spelled out in the charter.
>>> We already have people that are active on both WGs, and the chairs of
>>> both working groups contact each other when needed.
>>>
>>>
>>>
>>>> 2. Future of Identity Chaining: With
>>>> `draft-ietf-oauth-identity-chaining` nearing completion, does the new
>>>> charter keep the door open for further Workload Identity standardization
>>>> and chaining extensions, or is this particular scope considered complete
>>>> for now?
>>>>
>>>> No, this does not close the door for future work on this area, if it
>>> makes sense for that work to be at the OAuth WG.
>>>
>>>
>>>
>>>> 3. Alignment with W3C VCDM 2.0: Since our work on SD-JWT VC is heavily
>>>> tied to ecosystems like the EU Digital Identity Wallet, should the charter
>>>> explicitly mention ongoing alignment and maintenance regarding the W3C VCDM
>>>> 2.0 specifications?
>>>>
>>>> The chairs have close contacts with the EU Wallets initiative with
>>> frequent meetings to keep them up to date on the progress we are making on
>>> this front.
>>> Why do you think we need to explicitly mention the W3C VCDM 2.0
>>> specifications?
>>>
>>> Regards,
>>>  Rifaat
>>>
>>>
>>>
>>>> Looking forward to the discussion.
>>>>
>>>> Best regards,
>>>> Elazar.
>>>>
>>>> ‫בתאריך יום ה׳, 7 במאי 2026 ב-19:27 מאת ‪Rifaat Shekh-Yusef‬‏ <‪
>>>> [email protected]‬‏>:‬
>>>>
>>>>> All,
>>>>>
>>>>> The OAuth WG chairs and the Security AD (Deb) have been collaborating
>>>>> on a proposal to recharter the OAuth WG.
>>>>> https://datatracker.ietf.org/doc/charter-ietf-oauth/
>>>>>
>>>>> Deb put this on the agenda of the IESG telechat for *My 21st*.
>>>>>
>>>>> Please, take a look and let us know if you have any comments.
>>>>>
>>>>> Regards,
>>>>>  Rifaat & Hannes
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OAuth mailing list -- [email protected]
>>>>> To unsubscribe send an email to [email protected]
>>>>>
>>>> _______________________________________________ OAuth mailing list --
>>> [email protected] To unsubscribe send an email to [email protected]
>>>
>>> _______________________________________________
> OAuth mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>
_______________________________________________
OAuth mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to