I have added my initial comments to the Wiki (https://community.icann.org/x/WwAFBQ) and include them here for your convenience. I will develop this into a statement (likely complete mid-day Thursday) that I will submit on my own behalf. If anyone supports this, I will be glad to include others. Alan ==================== I have not had the time to do a very thorough analysis but I do have several comments based on my review of the document and related to Holly's comments. I agree that this work to date is a creature of the BC/IPC and the parties that have commented to date. I APPLAUD that. I an many of us have been saying for several months (certainly starting before the Abu Dhabi meeting) that the accreditation model is an absolute key to moving forward. In fact, some of you may recall that in answer to the Board's question to us about do we have any suggestions to mitigate the problems associated with GDPR, the ALAC answered YES. The answer was that we need to stop working in slow steps and start parallel efforts on all aspects of the problem, the most important being the accreditation model. We were ignored. The BC/IPC have done something and put it down on paper. Bravo! Holly has pointed out that we can only collect data that we need. Correct. But my interpretation of our mission is to provide a reliable trusted DNS. To do that we MUST ensure that those who help ensure that the DNS meets that target have access to the tools (and data) that they need to do their jobs. That IS our business!!! If we do not ensure that necessary data is collected, it can never be used by those with legitimate need. We need to state that clearly. Ultimately if I understand the process, it may be the courts that decide whether our rationale for collecting data is sufficiently strong to out-weight the privacy aspects. That brings us to the accreditation model, since WHO can get access to WHAT data is critical to us making that argument. I think this model is a great start. I see problems with it, but we need to start the discussion somewhere. What do I see as problematic? * I think their provision to give access to "All users", while it would be nice for me personally, is more than a bit loose. We are going to need something MUCH stronger to grant access over and above the other accredited channels. * I agree with Holly that a single tier is not sufficient. There should be more granularity based on the uses. Perhaps we could start with this single and improve later, but my preference would be to use the use-cases we already built (and how we spent the first many months of this project) to provide a few tiers. * I look forward to the work that the document says the SSAC is doing regarding credentialization. * I STRONGLY support the <https://community.icann.org/download/attachments/84213851/APWG-GDPR-Accreditationplancomments-5April2018-0001.pdf?version=1&modificationDate=1524004586320&api=v2>comments from the Anti-Phishing Working Group (APWG). These comments make the model more effective and implementable. At 17/04/2018 04:06 PM, Alan Greenberg wrote:

I said *I* would try to add some comments tonight. Add to what you had said (on Wiki now that it is there). Still hope to but time running out.

To be clear, it IS *OUR* job to make sure that the DNS is safe and reliable. That is our only job! Law enforcement, or anyone else that can justify their need can get data only if we first collect it.

But that is not relevant to the accreditation model as Scott says.

At 17/04/2018 07:13 AM, Holly Raiche wrote:

...

Hi Alan

When you say ?we can add some comments tonight? my questions are * what comments? * and add to what? There is no link on the policy page to the accreditation model.

And next - I?d have to differ on your views. Article 29 starts from basic privacy principles - as followed in over 100 countries around the globe with data protection regimes. First basic principle: only collect personal information that is necessary to do YOUR job. You don?t collect information you don?t need. Next principle: You tell the data subject why you need their personal information, and the circumstances in which that information will be shared. It is to THAT limited use and disclosure that people consent to giving their personal information. And it is in those LIMITED circumstances in which personal information is disclosed. The fact that others may want access is irrelevant.

So that right away limits who gets information - and in limited circumstances only.

In almost every data protection regime, law enforcement will be a circumstance in which personal information is made available. But again - not blanket permission, but in circumstances where there is a reasonable chance that illegal/harmful activity is or has taken place. And a further read suggests that access would also be given to Government/government authorised authorities as well. Two examples in Australia would be ASIC, our prudential regulator, and the ACCC, our competition and consumer watchdog. So the category of law enforcement probably extends to other regulatory/enforcement agencies - again, in situations where harm is suspected/occuring.

So - far from having an outstanding lack of knowledge about the Internet and ICANN, Article 29 know exactly what is going on. And it is what now amounts to unlawful release of personal information - and not just in the EU.

If there is to be any even slightly middle ground, it will be close to the model that Scott has already tested - access is possible, but to specific, accredited people and for specific situations. And it is in THAT context that the DNS will be protected. (and, as the cyber security advisor to our Prime Minister said, when asked how to make the Internet safe, replied that it isn?t possible to make it safe, only safer.)

Holly

On 17 Apr 2018, at 4:58 pm, Alan Greenberg <<mailto:alan.greenberg@mcgill.ca>alan.greenberg@mcgill.ca > wrote:

...

Holly I have been travelling and am ties up in meetings all day (Brussels time), but I hope to add some comments later tonight.

I caution care in interpreting the Article 29 letter. Despite some valid concerns, it shows an astounding lack of knowledge of the Internet and ICANN's mission. As a prime example is their advice to just focus on our own business and not that of others such as law enforcement or those combatting cyber issues. Our mission is to protect the DNS and that includes making it reliable and safe. But we do not have the ability to do that ourselves and thus must provide the tools for others to do that. If we fail, we are NOT carrying out our mission.

This will get more interesting in coming weeks.

Alan

At 16/04/2018 07:58 PM, Holly Raiche wrote:

...

Folks

We are fast running out of time to develop any kind of response to the accreditation model, even though the deadline for a response has been extended to this Friday. And in the interim, ICANN has received advice from Article 29 on the Interim model. (Article 29 is an advisory Group made up of a data protection authority from each EU member state). Their letter to ICANN, and Marby?s response, are on the home page of ICANN - and for those interested in the issue, I highly recommend reading both. Clearly, the implications of the Article 29 letter are that the Interim Model that we did comment on still raises concerns with them. Those concerns fall under the headings: * Breadth of purpose - saying the proposed purposes are too widely drawn * the link of purpose to processing - again, because the Whois data has been used does not qualify it as a purpose * publication of the data must be linked to the original (and narrowly defined) purpose * any access should be limited - not blanket access * discussion about the length of retention of data * discussion about the transfer of data * Accreditation (particularly important in this context) - should only be for legitimate purpose, limited to the original purpose, not blanket access, and under limited conditions

Below, I have copied in an email from Scott Hollenbeck, a long standing member of ICANN and one of those involved in the development of the RDAP (access protocol that would allow gated access to registration data) - simply because he has been involved in this issue for a long time and shows he has already worked on a technical solution to this issue - how access to data could work under GDPR.

I will try to attend as much as possible of the capacity building webinar on this issue, but have been scheduled to attend an all-day course in the Sydney CBD so may have to miss some of the discussion. I imagine Tom will be very up to date on these issues, but I would like to have been attending the whole of the webinar myself

In any case, if at all possible, we should be saying something. Quite apart from the original contribution from the IPC/BC model, the NCSG and Registrars have also made comments (largely challenging the IPC/BC model). I hope we are the one constituency that doesn?t make comments - although I realise that agreement on what to say will be difficult at the best of times.

And if it isn?t too late - maybe put this issue on the ALAC policy page - and with it, links to the Article 29 letter, Marby?s response, the registrars? response and the NCSG response (and any others I have missed - I have copies if that helps)

Holly

...

...

> >Hi all. > >After reading the Article 29 WP letter to ICANN >(<awbs://www.icann.org/en/system/files/correspondence/jelinek-to-marby-> >awbs://www.icann.org/en/system/files/correspondence/jelinek-to-marby- >11apr18-en.pdf), I started envisioning what process and system could >achieve GDPR compliance. What I came to is a token-based system, which >would work like this: >- Every request is analyzed by a human at an "RDS Clearinghouse". Each >request can be for a single data element (like "owner of >domain X") or to >multiple data elements (like "domains owned by the same owner of domain >X"), but requests for multiple data elements are only foreseen to be >processed by contracted parties with "Search WHOIS" contract >requirements. >- Clearinghouse issues a token with query parameters, data elements >authorized for response, identity of authorized party, reason for >authorization, validity (probably in the order of days), also informing >which endpoint to go to. >- Authorized party uses that token to access that endpoint, >managed by the >party with most data about that element (usually a registrar). > >Note that is not a replacement for credentialing; credentials >would still >be necessary to get tokens. This is also orthogonal to discussions like >which use cases are legitimate or not, GDPR-compliant or not etc.; it's >just a more granular approach to authorization that looks more >inline with >privacy-oriented guidelines including but not limited to GDPR.

Rubens, at a high level you just described how OpenID and OAuth work, except for the "Every request is analyzed by a human" part.

Scott,

I believe you are right, although most OAuth models I saw were not that granular to the point of saying "example.TLD, owner, e-mail address, valid until April 20 2018". That's not an OAuth limitation though, just common usage, and it probably could be made to work like this. And some level of asynchronous communications could even make way for a quick look human analysis.

Rubens

I have this very model, with human involvement, up and running right now as part of the gTLD RDAP Pilot. All of the attributes you mentioned can be encoded as OAuth claims. The model is described in an Internet-Draft that I first wrote in 2015. Just search for ?draft Hollenbeck RDAP OpenID? using your favorite search engine. Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Microsoft-Exchange-Diagnostics:

1;YTOPR01MB0396;27:E+eV2ZOSHNubNtVECuy/VM/L0xhmfXbZcuSQZbzTXSXqae7LUG8EA97rPf+2WwLbTntAgo7cpRKp5nw8d8T6ebvY8Vy/iFaEy7AJW3LXmiMa8o52Hh4X6vuxt8b9Dc+H X-Microsoft-Antispam-Message-Info:

mbIfSQV0+eZC39ENiUAspKdQkF0+JSTZwhHzgGzIJcnMTMtI1YFjxfAYUWl5x8Xsb0hCkvQc7VzC4AbTCpUbUGRQU6RcE5UUx1wEGk3WXKVvz27yXSf3osNfP8UwExDIUIL1xB46d04FaxAvbFZx3V7Y2ZVoGnlOF9jsJrXvgUSoJT4DsABV8smZdk4vNHRjunBYHmX/Uyc08KK5W5zM0fDRY7NEhuy8gp+3GzVz4A/5TqlICRbEqPiDy4iZWnOAaPiE4ELCAfZPmhMusibjqJno9CcaR04f9ovSYOL+LGRVBVtQ1M9b/SxTHtnAjl+ENjnGKfOC/j1+ryUspB55Z6d0P1zabIZYAwo87l1z4xMWzu3tmPExSwczo7sS129laKLyfrpQKQl6gt36u0ia78/RPd+UH+swgobcmXrp+pNPCIV60r6mKu9CLHxwfljZtI62wcXHykdzvBpIoyKjBQ==

_______________________________________________ ALAC mailing list <mailto:ALAC@atlarge-lists.icann.org>ALAC@atlarge-lists.icann.org https://atlarge-lists.icann.org/mailman/listinfo/alac

At-Large Online: http://www.atlarge.icann.org ALAC Working Wiki: https://community.icann.org/display/atlarge/At-Large+Advisory+Committee+(ALA... ) Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Microsoft-Exchange-Diagnostics:

1;YTOPR01MB0396;27:h8Wpez2dvY2XET3Ia5h4qAquPuwzm9VE+f0o9jAx4u4JqdcaeoFZrcLq85n2mYD3F4D2JBZ3d82Dv7LQjy5P6IxhefwMl889OIqCzWY/L0tSlen3d//E+BIL29ZiuFwf X-Microsoft-Antispam-Message-Info:

hY7ED33x23ZkuHwojyFbo9fT4+q8fh7UeaasB8GnDHvh0BflWaNXsKsfMieZOnBdYbD+6BGXeHDMiXHJ4CDj98J676LIfAnkqt5z4g81tEHWJeDXO46Rto11D8dDGA4rxI/gjrCwmWqpqYSFxjr5NSqyxKmG3FmUIfJkbMQ2MCGUw+UiodOO00BPDdel1hHHnTKdSFJh8jGtpFMe8Lki/O10iRkZbOfd3AjshkUl4bg=

_______________________________________________ ALAC mailing list ALAC@atlarge-lists.icann.org https://atlarge-lists.icann.org/mailman/listinfo/alac

At-Large Online: http://www.atlarge.icann.org ALAC Working Wiki: https://community.icann.org/display/atlarge/At-Large+Advisory+Committee+(ALA...)