Re: [OAUTH-WG] OAuth security topics

[Neil Madden <[e-mail]>]

Hi Torsten,

Apologies for the late reply. Suggested text:

----
4.x Client Impersonation of Resource Owner

Resource servers may make access control decisions based on the identity of the resource owner as communicated in the "sub" claim returned by the authorization server in a token introspection response [RFC 7662] or other mechanism. If a client is able to choose its own client_id during registration with the authorization server, then there is a risk that it can register with the same "sub" value as a privileged user. A subsequent access token obtained under the client_credentials grant may be mistaken as an access token authorized by the privileged user if the resource server does not perform additional checks.

4.1.x Proposed Countermeasures

Authorization servers SHOULD NOT allow clients to influence their client_id or "sub" value or any other claim that might cause confusion with a genuine resource owner. Where this cannot be avoided, authorization servers MUST provide another means for the resource server to distinguish between access tokens authorized by a resource owner from access tokens authorized by the client itself.
----

Open to revisions/suggestions.

-- Neil


> On 8 May 2019, at 13:15, Torsten Lodderstedt <[e-mail]> wrote:
> 
> You are right. May I ask you to propose text (threat description, advice, ...) to be included in the BCP? 
> 
>> On 7. May 2019, at 13:23, Neil Madden <[e-mail]> wrote:
>> 
>> The same issue applies to token introspection responses though.
>> 
>>> On 7 May 2019, at 12:21, Torsten Lodderstedt <[e-mail]> wrote:
>>> 
>>> Hi Neil,
>>> 
>>> you raised a very important issue but the Security BCP does not talk about JWT content like the Access Token JWT BCP.
>>> 
>>> I therefore think the security advise should go into the security considerations section of the Access Token JWT BCP.
>>> 
>>> best regards,
>>> Torsten.
>>> 
>>> Am 07.05.2019 um 11:18 schrieb Neil Madden <[e-mail]>:
>>> 
>>>> Is it too late to add the observation below to the OAuth security topics BCP draft?
>>>> 
>>>> 
>>>>> Begin forwarded message:
>>>>> 
>>>>> From: Neil Madden <[e-mail]>
>>>>> Subject: Re: [OAUTH-WG] draft-bertocci-oauth-access-token-jwt-00
>>>>> Date: 7 May 2019 at 09:37:29 BST
>>>>> To: Vittorio Bertocci <Vittorio=[e-mail]>
>>>>> Cc: Hans Zandbelt <[e-mail]>, Karl McGuinness <[e-mail]>, John Bradley <[e-mail]>, IETF oauth WG <[e-mail]>
>>>>> 
>>>>> I wasn’t at IIW so I may be missing some context.
>>>>> 
>>>>> There is a potential security issue if the client_id is used as the “sub” for an AT obtained via client_credentials. If the client can register itself with a self-chosen client_id then it may deliberately chose a client_id that matches the user name of a privileged user. An RS that just blindly looks at the “sub” claim may then erroneously let the client perform privileged actions.
>>>>> 
>>>>> Is this the context of the discussion?
>>>>> 
>>>>> Given that there are a lot of RSes in existence, many of which are probably just looking at the “sub” claim to identify the resource owner, I think the onus is on the AS to ensure that no such ambiguity can arise in the first place by ensuring that the space of “sub” values for client credentials is disjoint with the space for genuine users or by disallowing the client_credentials grant altogether.
>>>>> 
>>>>> This issue already arises in token introspection though, so maybe ought to be mentioned in the OAuth security topics draft rather than specific to the JWT AT draft?
>>>>> 
>>>>> — Neil
>>>>> 
>>>>>> On 6 May 2019, at 18:32, Vittorio Bertocci <Vittorio=[e-mail]> wrote:
>>>>>> 
>>>>>> Hi all,
>>>>>> thanks for your patience during this month long hiatus, and thanks for the comments.
>>>>>> Last week at IIW I had the opportunity to discuss this with many of the people on the list. Here's a summary of where the discussion landaed on the subject of the sub (pun intended).
>>>>>> 
>>>>>> - It seems that RFC 7519 has pretty clear language on the use of sub- I didn't check it back when we started the discussion. I do agree with the people saying that we shouldn't violate existing specifications, hence it looks like we do need to have sub also in the case in which we have app-only tokens carrying claims about the app itself. I understand this will cause some implementation to break, but unfortunately that's intrinsic in the process of coming up with a common approach and standards compliance is probably one of the strongest reasons to tolerate that.
>>>>>> - That said, I am strongly committed to preserve existing functionality. One of the main reasons that was brought up for including sub only for user based ATs was to use it as heuristic for telling those tokens apart from app-only ones. To that end, Karl MCGuinness suggested that we include grant_type as a return claim, which the RS could use to the same effect. I find the proposal very clever, and the people at IIW thought so as well. What you think?
>>>>>> Note, John Bradley observed that in some cases this might lead to ambiguous results if an extension grant is used (say an assertion profile) but that seems like a relatively fringe occurrence.
>>>>>> 
>>>>>> On Thu, Apr 4, 2019 at 1:11 PM Hans Zandbelt <[e-mail]> wrote:
>>>>>> I also meant to say that (of course) the JWT standard doesn't say that the JWT is supposed to be about the client or about the resource owner, hence both are valid
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> On Thu, Apr 4, 2019 at 10:09 PM Mike Jones <[e-mail]> wrote:
>>>>>> I get that existing practice is likely to be all over the map, but if we’re to create a JWT access token standard, it’s reasonable to require that the claims usage comply with the JWT standards.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>>                                                                -- Mike
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> From: Hans Zandbelt <[e-mail]> 
>>>>>> Sent: Thursday, April 4, 2019 12:59 PM
>>>>>> To: Mike Jones <[e-mail]>
>>>>>> Cc: George Fletcher <gffletch=[e-mail]>; Vittorio Bertocci <Vittorio=[e-mail]>; IETF oauth WG <[e-mail]>
>>>>>> Subject: Re: [OAUTH-WG] draft-bertocci-oauth-access-token-jwt-00
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> the definition of RFC 7519 is actually "petitio principii" and a lot of deployments put claims about the Resource Owner in the access token (as a Resource Server implementer I've suffered a lot from this)
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Thu, Apr 4, 2019 at 9:48 PM Mike Jones <[e-mail]> wrote:
>>>>>> 
>>>>>> I agree with George that the subject of a token must be the principal of that token.  That what the JWT “sub” claim is for.  Indeed, the first sentence of the “sub” definition at https://tools.ietf.org/html/rfc7519#section-4.1.2 is:
>>>>>> 
>>>>>> The "sub" (subject) claim identifies the principal that is the subject of the JWT.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> If an access token uses “sub”, its usage must comply with the definition from RFC 7519.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>>                                                                -- Mike
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> From: OAuth <[e-mail]> On Behalf Of George Fletcher
>>>>>> Sent: Thursday, April 4, 2019 8:51 AM
>>>>>> To: Hans Zandbelt <[e-mail]>
>>>>>> Cc: Vittorio Bertocci <Vittorio=[e-mail]>; IETF oauth WG <[e-mail]>
>>>>>> Subject: Re: [OAUTH-WG] draft-bertocci-oauth-access-token-jwt-00
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> The more I think about this the more I land in the "No" camp.
>>>>>> 
>>>>>> The subject of a token should be the principal of that token. It shouldn't matter whether that is a machine, a user, or a device. Trying to separate out "humans" as a special class will just make things more complicated. If we need a claim to identify the subject is a "human" then why not just add that. This doesn't break anything and makes it easy for people to detect this case in those use cases where it's required.
>>>>>> 
>>>>>> Thanks,
>>>>>> George
>>>>>> 
>>>>>> On 4/3/19 4:56 PM, Hans Zandbelt wrote:
>>>>>> 
>>>>>> I will argue that in a way such deployments are already broken e.g. in the typical use case of onboarding client accounts in the same directory/OU/namespace as user accounts and we don't need to cater for that.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Wed, Apr 3, 2019 at 10:48 PM George Fletcher <[e-mail]> wrote:
>>>>>> 
>>>>>> I agree that this will break a lot of existing flows... especially those using any form of the client_credentials flow. In that sense I'm not completely on board yet :)
>>>>>> 
>>>>>> On 3/26/19 12:56 PM, Hans Zandbelt wrote:
>>>>>> 
>>>>>> great summary! this will hurt quite a few existing m2m deployments but I do like the rigidness of it all: it is very explicit, cannot misinterpreted and thus prevents failure (which is really what Dominick is after); I'm on board
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Tue, Mar 26, 2019 at 5:49 PM Vittorio Bertocci <Vittorio=[e-mail]> wrote:
>>>>>> 
>>>>>> thank you Steinar and everyone else for the comments on this!
>>>>>> 
>>>>>> To summarize the situation so far: Dominick, Steinar, Rob, David, Nov, Bertrand recommend using sub only for users. Martin would like to have the sub for app only flows as well. Hans is neutral.
>>>>>> 
>>>>>> That does sound like the sub as user has more consensus, tho before changing it I'd wait for the people currently at IETF104 to have more time to comment as well.
>>>>>> 
>>>>>> Clarification. If the goal is to be able to apply the logic "if there's a sub, it's a user flow", we have to explicitly disallow (MUST NOT) the use of sub when that's not the case. Are all OK with it?
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Dave, the suggestion of having explicit typing for app only vs user only is interesting! For the purpose of putting together an interoperable profile, tho, I would suggest we table it for v1 in the interest of getting to something easy to adopt (hence with small delta vs existing implementations) faster.     
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Tue, Mar 26, 2019 at 1:40 AM Steinar Noem <[e-mail]> wrote:
>>>>>> 
>>>>>> Hi Vittorio, we  (the national federation-gateway for the health services in norway - "HelseID")  think his is a really valuable initiative!
>>>>>> 
>>>>>> We also agree with Dominick concerning definition of the "sub" claim.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> <mvh>Steinar</mvh>
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> tir. 26. mar. 2019 kl. 07:25 skrev Dominick Baier <[e-mail]>:
>>>>>> 
>>>>>> From an access token consumer (aka API) developer point of view, I prefer this logic
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> "If sub is present - client acts on behalf of a user, if not - not."
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Anything more complicated has a potential of going wrong.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On 26. March 2019 at 01:34:53, Nov Matake ([e-mail]) wrote:
>>>>>> 
>>>>>> Hi Vittorio,
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Yeah, I’m concerning user & client ids collision.
>>>>>> 
>>>>>> I haven’t seen such implementations, but user-select username as sub, or incremental integer as sub & client_id will be easily collide.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> If you can enforce collision resistant IDs between user & client instances, it’ll works fine. I feel its overkill though.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Sent from my iPhone
>>>>>> 
>>>>>> 
>>>>>> On Mar 26, 2019, at 8:51, Vittorio Bertocci <[e-mail]> wrote:
>>>>>> 
>>>>>> Hey Nov, Dominick, Hans-
>>>>>> 
>>>>>> thanks for the comments. That was an area I was expecting to cause more discussion, and I am glad we are having this opportunity to clarify.
>>>>>> 
>>>>>> The current language in the draft traces the etymology of sub to OpenID Connect core, hence Dominick observation is on point. However in the description I express something in line with 7519, which was in fact my intent.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> The idea was to provide an identifier of the calling subject that is guaranteed to be present in all cases- this would allow an SDK developer to use the same code for things like lookups and membership checks regardless of the nature of the caller (user in a delegated case, app in app-only grants). The information to discriminate between user and app callers is always available in the token (say, the caller is a user if sub!=client_id, where client_id is always guaranteed to be present as well) hence there's no loss in expressive power, should that difference be relevant for the resource server.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Dominick, Hans- I probably missed the security issue you guys are thinking of in this case. Of course, if this would introduce a risk I completely agree it should be changed- I'd just like to understand better the problem. Could you expand it in a scenario/use case to clarify the risk?
>>>>>> 
>>>>>> Nov- playing this back: is the concern that a user and a client might have the same identifier within an IDP? When using collision resistant IDs, as it is usually the case, that seems to be a remote possibility- did you stumble in such scenario in production?
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Thanks
>>>>>> 
>>>>>> V. 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Mon, Mar 25, 2019 at 7:44 AM Hans Zandbelt <[e-mail]> wrote:
>>>>>> 
>>>>>> I believe there are plenty of OAuth 2.0 only use cases out there... but nevertheless I agree with the potential confusion and thus security problems arising from that (though one may argue the semantics are the same).
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Mon, Mar 25, 2019 at 3:39 PM Dominick Baier <[e-mail]> wrote:
>>>>>> 
>>>>>> Yes I know - and I think in hindsight it was a mistake to use the same claim type for multiple semantics.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> All the “this is OIDC not OAuth” arguments are making things more complicated than they need to be - in my experience almost no-one (that I know) does OIDC only - nor OAuth only. They always combine it.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> In reality this leads to potential security problems - this spec has the potential to rectify the situation.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Dominick
>>>>>> 
>>>>>> On 25. March 2019 at 14:58:56, Hans Zandbelt ([e-mail]) wrote:
>>>>>> 
>>>>>> Without agreeing or disagreeing: OIDC does not apply here since it is not OAuth and an access token is not an id_token.
>>>>>> 
>>>>>> The JWT spec says in https://tools.ietf.org/html/rfc7519#section-4.1.2:
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> "The "sub" (subject) claim identifies the principal that is the
>>>>>> 
>>>>>>   subject of the JWT.  The claims in a JWT are normally statements
>>>>>> 
>>>>>>   about the subject.  The subject value MUST either be scoped to be
>>>>>> 
>>>>>>   locally unique in the context of the issuer or be globally unique.
>>>>>> 
>>>>>>   The processing of this claim is generally application specific"
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> which kind of spells "client" in case of the client credentials grant but I also do worry about Resource Servers thinking/acting only in terms of users
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Hans.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Mon, Mar 25, 2019 at 2:41 PM Dominick Baier <[e-mail]> wrote:
>>>>>> 
>>>>>> IMHO the sub claim should always refer to the user - and nothing else.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> OIDC says:
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> "Subject - Identifier for the End-User at the Issuer."
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> client_id should be used to identify clients.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> cheers
>>>>>> 
>>>>>> Dominick
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On 25.. March 2019 at 05:13:03, Nov Matake ([e-mail]) wrote:
>>>>>> 
>>>>>> Hi Vittorio,
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Thanks for the good starting point of standardizing JWT-ized AT.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> One feedback.
>>>>>> 
>>>>>> The “sub” claim can include 2 types of identifier, end-user and client, in this spec.
>>>>>> 
>>>>>> It requires those 2 types of identifiers to be unique each other in the IdP context.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> I prefer omitting “sub” claim in 2-legged context, so that no such constraint needed.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> thanks
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> nov
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> On Mar 25, 2019, at 8:29, Vittorio Bertocci <vittorio.bertocci=[e-mail]> wrote:
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Dear all,
>>>>>> 
>>>>>> I just submitted a draft describing a JWT profile for OAuth 2.0 access tokens. You can find it in https://datatracker.ietf.org/doc/draft-bertocci-oauth-access-token-jwt/.
>>>>>> 
>>>>>> I have a slot to discuss this tomorrow at IETF 104 (I'll be presenting remotely). I look forward for your comments!
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Here's just a bit of backstory, in case you are interested in how this doc came to be. The trajectory it followed is somewhat unusual.
>>>>>> 
>>>>>> 	• Despite OAuth2 not requiring any specific format for ATs, through the years I have come across multiple proprietary solution using JWT for their access token. The intent and scenarios addressed by those solutions are mostly the same across vendors, but the syntax and interpretations in the implementations are different enough to prevent developers from reusing code and skills when moving from product to product.
>>>>>> 	• I asked several individuals from key products and services to share with me concrete examples of their JWT access tokens (THANK YOU Dominick Baier (IdentityServer), Brian Campbell (PingIdentity), Daniel Dobalian (Microsoft), Karl Guinness (Okta) for the tokens and explanations!).
>>>>>> I studied and compared all those instances, identifying commonalities and differences. 
>>>>>> 	• I put together a presentation summarizing my findings and suggesting a rough interoperable profile (slides: https://sec.uni-stuttgart.de/_media/events/osw2019/slides/bertocci_-_a_jwt_profile_for_ats.pptx ) - got early feedback from Filip Skokan on it. Thx Filip!
>>>>>> 	• The presentation was followed up by 1.5 hours of unconference discussion, which was incredibly valuable to get tight-loop feedback and incorporate new ideas. John Bradley, Brian Campbell Vladimir Dzhuvinov, Torsten Lodderstedt, Nat Sakimura, Hannes Tschofenig were all there and contributed generously to the discussion. Thank you!!!
>>>>>> Note: if you were at OSW2019, participated in the discussion and didn't get credited in the draft, my apologies: please send me a note and I'll make things right at the next update.
>>>>>> 	• On my flight back I did my best to incorporate all the ideas and feedback in a draft, which will be discussed at IETF104 tomorrow. Rifaat, Hannes and above all Brian were all super helpful in negotiating the mysterious syntax of the RFC format and submission process.
>>>>>> I was blown away by the availability, involvement and willingness to invest time to get things right that everyone demonstrated in the process. This is an amazing community. 
>>>>>> 
>>>>>> V.
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> [e-mail]
>>>>>> 
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> [e-mail]
>>>>>> 
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> Vennlig hilsen
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> Steinar Noem
>>>>>> 
>>>>>> Partner Udelt AS
>>>>>> 
>>>>>> Systemutvikler
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> | [e-mail] | [e-mail]  | [phone] | www.udelt.no | 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> [e-mail]
>>>>>> 
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> [e-mail]
>>>>>> 
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> [e-mail]
>>>>>> 
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> -- 
>>>>>> [e-mail]
>>>>>> ZmartZone IAM - www.zmartzone.eu
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>> 
>>>> 
>> 
>