Ryan, I definitely agree there is a security risk with 3.2.2.4.1 and any other validation method entirely dependent on a “trust us” model. I did see the specifics Jeremy shared about the problems with method 1 in an earlier thread. You mention in your first reply that “To date, Entrust has not provided any of the requested details about its use of 3.2.2.4.1, the prevalence, and the potential impact” and then later state “the security risk posed by the arbitrariness allowed, which Entrust has specifically demonstrated their systems are vulnerable to, presents an unacceptably large risk to our users.”
Did I miss a post by Entrust where they specifically provided this security risk detail? Bruce do you have this info? It sounds like all of us agree this risk needs to be mitigated and that some CAs are more capable than others to react to a change in validation method used. I’m trying to get a feel for how much of our ecosystem (either # CAs, % of customer base or % of certs) relies on method 1 and what impact or disruption is caused to customers (both individual and enterprise) if a significant part of the CA community can’t respond quickly and efficiently to the change. This may support a more deliberate implementation for the good of the entire ecosystem vs. just a few CAs or one Browser. Given Tim sent out Ballot 218 version 2 for discussion with an implementation date of August 2018 it sounds like Google would vote “no” if it went to vote. I suppose how each CA votes would be illustrative of how much of the ecosystem is ready for an August implementation. I’d really like to see the f2f meetings used to discuss these type of ballots and given #43 is a month away the timing seems appropriate. Discussion could continue for another month via email and teleconference and perhaps a better view of the business impact to customers could be assessed against the security risk of a graceful implementation. Thanks, Mike From: Ryan Sleevi [mailto:[email protected]] Sent: Monday, January 29, 2018 10:18 AM To: Doug Beattie <[email protected]> Cc: CA/Browser Forum Public Discussion List <[email protected]>; Mike Reilly (WDG) <[email protected]> Subject: Re: [cabfpub] Ballot 218 version 2: Remove validation methods #1 and #5 I don't think it's misleading - both GlobalSign and Let's Encrypt took steps to disable issuance to avoid introducing further risk. Both GlobalSign and Let's Encrypt than proposed future steps to ensure that the risks were mitigated, to various degrees of comprehensiveness. These steps were taken within hours and days, respectively, of becoming aware of issues, and it's a credit to both GlobalSign and Let's Encrypt the degree of responsiveness that was undertaken. I think that both CAs took proactive steps to cease issuance, and engage in the community, is reflective of both what is reasonable and what is exemplary, particularly given the challenges in disabling automated forms of issuance (compared to, say, manual forms, in which it is already routinely expected to engage with the customer-applicant) Compare and contrast that with Bruce's proposal, which provides for no mitigations for the many issues already discovered in the level of assurance provided by Entrust's use of 3.2.2.4.1, which presently allows for up to 3 year certificates to be issued (to be reduced to 825-days in March), and further proposes to allow the reuse of that validation information. To really emphasize this point, as it may be missed, the maximum 'damage' an improperly validated certificate validated under 3.2.2.4.10 by Let's Encrypt can do is measured by 90 days from its issuance, or, if we were to look at today, 2018-04-29. The maximum 'damage' an improperly validated certificate validated under 3.2.2.4.1 by Entrust, as proposed by Bruce, is measured as 2021-04-29 (if DV/OV), or 2020-04-29 (if EV) - if the move was made today, or 2021-05-28 (DV/OV) and 2020-11-03 (EV) if adopted by August. I don't think it's misleading to suggest that the responses to 3.2.2.4.9 and 3.2.2.4.10, by GlobalSign and Let's Encrypt, respectively, are reflective of positive committments to ensuring that users and site operators are secure and protected from improper validation - even when the risk is mitigatable - and the reactions to 3.2.2.4.5 and 3.2.2.4.1, for which both no mitigations exist, nor have the inherent security issues been acknowledged or responded to, are of a different kind and calibre. On Mon, Jan 29, 2018 at 1:03 PM, Doug Beattie <[email protected]<mailto:[email protected]>> wrote: Ryan, I think the term “disable” is a bit strong, or perhaps misleading in this context. Yes, Let’s Encrypt did was temporarily disabled method 10 within hours, but isn’t method 10 still being used to issue a large number of certificates, and isn’t renewal of domains that used this method still allowed? I wouldn’t call that “disabled”. From: Public [mailto:[email protected]<mailto:[email protected]>] On Behalf Of Ryan Sleevi via Public Sent: Monday, January 29, 2018 12:42 PM To: Mike Reilly (WDG) <[email protected]<mailto:[email protected]>> Cc: CA/Browser Forum Public Discussion List <[email protected]<mailto:[email protected]>> Subject: Re: [cabfpub] Ballot 218 version 2: Remove validation methods #1 and #5 As a further point of comparison regarding impact versus risk - CAs were successfully able to disable the use of methods 3.2.2.4.9 and 3.2.2.4.10 within hours of determining they may provide less than reliable information, despite such validation methods accounting for more than half of some CAs issuance. We think this more than demonstrates that security-conscious CAs can take steps to mitigate risk, and to do so transparently, in a rapid timeframe, and thus are concerned about CAs that present it as overly burdensome or difficult, given the risks. On Mon, Jan 29, 2018 at 12:29 PM, Ryan Sleevi <[email protected]<mailto:[email protected]>> wrote: Hi Mike, To be clear, we (Google) do not believe that August represents a suitable balancing of risk. As Bruce notes, we are examining moving sooner on this - the security risk posed by the arbitrariness allowed, which Entrust has specifically demonstrated their systems are vulnerable to, presents an unacceptably large risk to our users. From a technical perspective, the fact that validation methods (and their use) is entirely dependent on a 'trust us' approach based by CAs, as compared to say, requirements on the use of signature algorithms, certificate profiles, or revocation actions, further presents complications, in that it indirectly ties browser security policy to a majority of CAs agreeing. This sort of product-level decision remains best dealt with at a product level, and to the extent the Forum can provide harmonized discussion, we do want to take advantage of it. We believe that the information shared - by Entrust (who believes it acceptable) and by DigiCert (who understandably expressed concerns with how it's historically used) - and the past discussions of and information shared on how CAs are using these methods, such as during our Berlin F2F last year, provide sufficient context as to the risk. With regard to CA practices, given CAs' disclosures in their CP/CPS, and the ready availability of methods .2 and .3 - which involve the same amount of validation effort (a Reliable Method of Communication, the same as used in 3.2.5), but use reliable information rather than unreliable information, more than sufficiently mitigate the risk posed. As a practical matter, Entrust's proposal effectively attempts to stifle action for another two months, with a further completely unacceptable proposal to continue to reuse the (insecurely validated) information. We believe that CAs can, and must, be able to respond more effectively, and more responsibly, then making a change by then. This includes, to be clear, the August date - which attempts to seek consensus in the Forum for what is already an upper-bound on the level of security that users expect and the level of control - and ability to mitigate misissuance - that site operators expect. On Mon, Jan 29, 2018 at 11:52 AM, Mike Reilly (WDG) <[email protected]<mailto:[email protected]>> wrote: Hi Bruce and Ryan. Based on my review of the ballot 218 discussion, I believe it would be beneficial to discuss this ballot at our next f2f meeting which is just about a month away. This is an important topic for all members of the ecosystem (customers, CAs, and Browsers) and the outcome will impact some CAs more that others. I’d prefer to see more deliberate planning toward mitigating this security risk, which the August date seams to support. From my understanding this discussion started with Jeremy’s proposal on 19 December, which was right before the time most were away on holiday break. So, it has been in discussion for just over a month now, but via email and part of one CABF call. I’m still a relatively new member of the CABF and see the f2f meetings as a great setting to discuss these larger issues. If there are elements of the discussion which are missing perhaps the best use of the mail list would be to centrally collect the concerns (ideally with some data from key stakeholders) in order to facilitate a productive f2f meeting in early March. I’ve not seen good data on actual security incidents related to this method nor have I seen data on how much members of our CA ecosystem rely on this method. I definitely believe 3.2.2.4.1 needs improvement or elimination for the security of customers. However, I would also like to see the risk mitigated in a deliberate vs. crisis mode fashion to minimize disruption to the ecosystem. Seems we could agree on landing a vote immediately after f2f #43. Thanks, Mike From: Public [mailto:[email protected]<mailto:[email protected]>] On Behalf Of Ryan Sleevi via Public Sent: Monday, January 29, 2018 7:21 AM To: Bruce Morton <[email protected]<mailto:[email protected]>>; CA/Browser Forum Public Discussion List <[email protected]<mailto:[email protected]>> Subject: Re: [cabfpub] Ballot 218 version 2: Remove validation methods #1 and #5 Bruce, To date, Entrust has not provided any of the requested details about its use of 3.2.2.4.1, the prevalence, and the potential impact. Given the lack of responsiveness to the issues, which were raised over a month ago, and have made no substantial progress to addressing or understanding the security considerations, it does not seem like another two months of discussion will be productive, particularly given the user-security risk. Given the lack of information and responsiveness to the issues, it is difficult to believe this is a good faith request, and not simply yet another attempt to stall conversation in the Forum. We are certainly sensitive to the potential impact, based on available data - but CAs, such as Entrust's, unwillingness to contribute effectively to that conversation, or ability to demonstrate an awareness of and sensitivity to the security risks presented to the ecosystem by the current practices, leads to the unavoidable conclusion that there is not much productive progress. We value the Forum for its excellent opportunities to discuss proposals, share information, and assess the impact to the security of users about various proposals - including the potential challenges faced by site operators. Yet that does not preclude taking necessary steps to protect the safety and security of users, to accurately report to users the level of assurance provided by a given certificate, and to take steps to reduce the risk from CAs whose validation practices are insufficient for the needs of the Web PKI. Thus, we do not believe further delays, given the direct lack of responsiveness to the matters, is either warranted or wise. On Mon, Jan 29, 2018 at 10:01 AM, Bruce Morton via Public <[email protected]<mailto:[email protected]>> wrote: On the CA/Browser Teleconference last Thursday, the members discussed pending Ballot 218, which would eliminate domain validation method 1 (WhoIs lookup, BR 3.2.2.4.1) as of August, 2018. Google indicated it was not satisfied with an August 2018 implementation date, and might impose a March 2018 date on its own (through the Google root program) ending the ability to use Method 1 for domain validation as of that date. We would like to ask Forum members, including Google, for a little more time to discuss this issue, including possible amendments to Ballot 218 that might satisfy everyone’s concerns. Some possible ideas for amending Method 1 (which we can develop further in our next meeting) could include the following: * Strengthen Method 1 by adding more details match the Applicant and the domain Registrant using name and a unique identifier. * Require the CA also to send a notice of domain validation to the Registrant, allowing the Registrant to have the certificate revoked if the validation is not authorized. * Eliminating Method 1 for DV and OV domain validation, but allowing Method 1 to be used for EV validation. The EV validation process already includes two steps to confirm the authority of the Applicant Representative to order the certificate – including a call to a second person at the organization to confirm that the Applicant Representative has authority to request the certificate. (EVGL 11.8) * Ballot 218 will require revalidation of 100% of domains previously validated using Method 1 for issuance starting 1 August 2018 (if Google acts unilaterally to prohibit use of Method 1 by March, this revalidation deadline could be March 2018). We suggest Ballot 218 should allow reuse of domain validation data for the normal period allowed by BR 4.2.1, as is the Forum’s standard practice. This will allow CAs to change processes, implement/extend automation and train customers on alternative validation methods. * Finally, the elimination of Method 1 will have a significant impact not only on CAs (some of whom are Forum members, but many of whom are not and are unaware of this discussion), but more importantly on major certificate users including enterprises and governments who often prefer Method 1 for validation of their domains. Therefore, we ask for more time for both discussion and implementation so both CAs and website owners can have a more graceful transition to whatever new rules we ultimately adopt. The Forum will be meeting in approximately 5 weeks at our Face to Face meeting in Herndon – this is a complex topic which would benefit from a thorough discussion at that meeting to try to reach a sensible solution. We should continue discussion on this list, but let’s wait until the F2F meeting occurs to reach a final conclusion that can be implemented without unnecessary disruption to the security ecosystem. Thanks, Bruce. From: Public [mailto:[email protected]<mailto:[email protected]>] On Behalf Of Tim Hollebeek via Public Sent: January 22, 2018 4:31 PM To: CA/Browser Forum Public Discussion List <[email protected]<mailto:[email protected]>> Subject: [EXTERNAL][cabfpub] Ballot 218 version 2: Remove validation methods #1 and #5 Ballot 218 version 2: Remove validation methods #1 and #5 Purpose of Ballot: Section 3.2.2.4 says that it “defines the permitted processes and procedures for validating the Applicant’s ownership or control of the domain.” Most of the validation methods actually do validate ownership and control, but two do not, and can be completed solely based on an applicant’s own assertions. Since these two validation methods do not meet the objectives of section 3.2.2.4, and are actively being used to avoid validating domain control or ownership, they should be removed, and the other methods that do validate domain control or ownership should be used. The following motion has been proposed by Tim Hollebeek of DigiCert and endorsed by Ryan Sleevi of Google and Rich Smith of Comodo. -- MOTION BEGINS – This ballot modifies the “Baseline Requirements for the Issuance and Management of Publicly-Trusted Certificates” as follows, based upon Version 1.5.4: In Section 1.6.1, in the definition of “Domain Contact”, after “in a DNS SOA record”, add “, or as obtained through direct contact with the Domain Name Registrar” In Section 3.2.2.4.1, add text at the end: “For certificates issued on or after August 1, 2018, this method SHALL NOT be used for validation, and completed validations using this method SHALL NOT be used for the issuance of certificates.” In Section 3.2.2.4.5, add text at the end: “For certificates issued on or after August 1, 2018, this method SHALL NOT be used for validation, and completed validations using this method SHALL NOT be used for the issuance of certificates.” After Section 3.2.2.4.10, add following two new subsections: “3.2.2.4.11 Any Other Method This method has been retired and MUST NOT be used. 3.2.2.4.12 Validating Applicant as a Domain Contact Confirming the Applicant's control over the FQDN by validating the Applicant is the Domain Contact. This method may only be used if the CA is also the Domain Name Registrar, or an Affiliate of the Registrar, of the Base Domain Name. Note: Once the FQDN has been validated using this method, the CA MAY also issue Certificates for other FQDNs that end with all the labels of the validated FQDN. This method is suitable for validating Wildcard Domain Names.“ In Section 4.2.1, after the paragraph that begins “After the change to any validation method”, add the following paragraph: “Validations completed using methods specified in Section 3.2.2.4.1 or Section 3.2.2.4.5 SHALL NOT be re-used on or after August 1, 2018.” -- MOTION ENDS – For the purposes of section 4.2.1, the new text added to 4.2.1 from this ballot is “specifically provided in a [this] ballot.” The procedure for approval of this ballot is as follows: Discussion (7+ days) Start Time: 2017-01-22 21:30:00 UTC End Time: Not Before 2017-01-29 21:30:00 UTC Vote for approval (7 days) Start Time: TBD End Time: TBD _______________________________________________ Public mailing list [email protected]<mailto:[email protected]> https://cabforum.org/mailman/listinfo/public<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcabforum.org%2Fmailman%2Flistinfo%2Fpublic&data=02%7C01%7CMike.Reilly%40microsoft.com%7C068849bea0824585f06008d5672c0305%7Cee3303d7fb734b0c8589bcd847f1c277%7C1%7C0%7C636528361105059910&sdata=6Mp6NGQn7Bs9wbOt2u7B2x6lVPPLH%2BfZyhOzwD22ELo%3D&reserved=0>
_______________________________________________ Public mailing list [email protected] https://cabforum.org/mailman/listinfo/public
