On Oct 14, 2022, at 9:23 AM, AJITOMI Daisuke <[email protected]> wrote:
> If you are not aware of the difference and insist, "I (as a Holder) have the 
> right to freely disclose the issuer's data (related to me) !!!!!!!!", then I 
> have to explain the difference at least.
> 
> Simply put, the difference is that an existing DMV doesn't provide a personal 
> data disclosure API (such as /userinfo), while JWP/SD-JWT does (at least, 
> SD-JWT does).
> 
> The former includes the case where the Issuer provides only a public key (for 
> example, via jwks_uri) to verify the signature of the issued data.
> A typical example is the digital COVID vaccination certificate in the US. And 
> I don't think that these cases are problematic.
> 
> On the other hand, the latter(JWP or SD-JWT) assumes that the Issuer will 
> provide a personal data disclosure API, which returns JWP or SD-JWT formatted 
> data respectively, as a service. In this case, to my understanding, the 
> Issuer is a personal data controller and has to fulfill various 
> responsibilities for personal data management (**See question2 mentioned in 
> my first post**) and the Holder cannot act as other than a proxy of the 
> Issuer.
> I said that if you are arguing that the holder does not have to act as a 
> proxy for the issuer, please tell me the rationale (**See the response to 
> David's post**).

Apologies Daisuke, I appreciate your effort to explain the difference but I do 
not quite understand as JWT, SD-JWT, and JWP are all data formats.

My existing process for getting a digital COVID credential (particularly SMART 
Health Cards) was to have my pharmacy, my insurance provider or my state 
government to vouch for vaccination data. This was a VC based on a FHIR 
document of vaccine records, e.g. I was vaccinated at this time at this 
organization with this vaccine purpose, variant/manufacturer, and the lot 
number.

It also includes identifying information such as my full name and date of 
birth, as well as values which could be used for multi-party correlation based 
on the record (if full name and birth date were somehow not enough to do so)

While having someone be able to get their own medical records like this is 
empowering, the primary use case in people’s minds was to prove sufficient 
vaccination for entry or participation in some activity. That did not rely on 
giving nearly as much personal data. In addition, some of the data and extra 
requirements of the system (e.g. to have someone show another form of 
photographic identification with their full name) existed because the QR code 
was just signed data and did not supply a way to show any additional 
cryptographic proof that I was the subject of the credential.

That lack of additional proof also meant there was no providence of the record 
- not only could the recipient share information, but there was no way to tell 
if there had been consent granted for them to do so. For example, there was an 
app meant to verify the data in the credential and to list the full name so 
that people could consume these credentials. However, if I scanned a COVID 
credential using the built-in camera app on my phone, it instead imported their 
health record and allowed me to present it in the future.

Different COVID credentials existed, with different capabilities. For SD-JWT or 
JWP as a data format to replace the JWT credential above and to solve these 
privacy concerns seems like a pretty clear data protection win.

Would I be correct in assuming the concern is not in the data formats 
themselves, but in the potential protocols being used to issue these 
credentials? For instance, that they are issued interactively (via out of scope 
protocols) and potentially refreshed (via an API), vs being a static 
‘vaccination certificate’ handed to the user in the form of a scannable image?

Or is the concern more that these types of data formats encourage the ability 
for the data itself to be manipulated in certain ways in general, e.g. by a 
holder agent to reduce the amount of information shared. The concern then is 
that manipulation would cause the software holding the credential to be 
considered a data processor with a relationship with the original issuing 
authority under certain regulations, rather than merely an agent under the 
user’s control?

-DW




_______________________________________________
jose mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/jose

Reply via email to