I wanted to elaborate on a piece of my last message, specifically around
issuance of certificates for uses that are not compatible with the
misissuance revocation requirements specified in the BRs.

We very commonly see that delayed revocation incidents are rationalized by
CAs as being required due to the impact of disruption of services that
would result from compliant revocation. Many different types of service are
cited as being in this "too big to revoke" category, ranging from banks to
airlines to post offices to government portals. This is sometimes described
as being "disruptive to the web ecosystem".

One thing that could be done to reduce this is to explicitly call out
certain uses that should not rely on certificates issued under the
authority of the BRs. There are already some prohibitions on use of such
certificates for life-critical systems, nuclear reactor control, etc.--the
same as commonly found in software licenses.

Rather than try to litigate and maintain a list of usage descriptors, I
propose instead letting CAs determine the suitability of WebPKI use.

Specifically, the BRs could be amended to indicate that it is misissuance
to issue a certificate to a subscriber that runs a service that cannot
endure revocation and can't replace certificates in 24 hours in "emergency,
overtime, disrupt other stuff" mode, or 5 days in "business as usual mode".
This would mean that even an otherwise perfectly-valid certificate would be
considered to be misissued if it is somehow discovered that a subscriber
operates a socially-critical system that cannot tolerate compliant
revocation, and the CA would then no longer be able to renew or otherwise
issue to that subscriber for that service until they became confident that
the subscriber was now "revocation-tolerant". Maybe.

I'm not totally satisfied with how I'm thinking about this yet, but I think
the kernel of the idea is sound on both core points:
- critical systems that cannot tolerate BR-compliant revocation should not
be issued certificates that are subject to the BRs
- the CAs should be responsible for ensuring that they are not issuing
certificates for inappropriate uses

This would build on whatever work CAs are currently doing to ensure that
customers are aware that WebPKI certificates are not suitable for use in
medical devices or reactor control systems, I assume.

Mike

-- 
You received this message because you are subscribed to the Google Groups 
"[email protected]" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/a/mozilla.org/d/msgid/dev-security-policy/CADQzZquh3LVMBzyQ9RM%3D9WSXYEMt8uyVgLt5cOH5EEDgtx%2B_LA%40mail.gmail.com.

Reply via email to