DEF CON 23 – Panel – Licensed to Pwn: Weaponization and Regulation of Security Research


It’s my pleasure. You’re getting
a real stack of attorneys today. We had alley in here and he was
a attorney and then a panel of a bunch of them this is going to
be a really interesting set about the weaponization and
regulation of security research. Something that is pretty
important to a lot of us here. I’m not going to introduce
everybody individually. I’ll probably let Jim take care of
that. So, let’s give our panel a big hand. (Applause)>>JIM
DENARO: Thank you for coming to our panel. This is licensed to
own. The weaponization of security research. We appreciate
you taking the time to explore this important issue with us.
This is a unique time for the entire information security
community. And shear why. The U.S. government is implementing
rules that could change the way that information security is
practiced in the United States. And more than that, it could
even affect the way that we talk about information security. The
way information is exchanged among us here. And if that
sounds like a big deal, it is. And that’s not an overstatement
at all of the situation that we are currently looking at.
Specifically, some new regulations are being proposed
by the Bureau of Industry and security, also known as BIS as
what we will refer to it here. That’s within the U.S.
Department of Commerce. Stated mission of BIS is to administer
and enforce dual use export controls on various
technologies. Such as body armor, bulletproof windshield
glass, encryption, most relevant to what we’re doing, and those
charges are now being expanded to cover some technologies that
we are seeing more often here in this room, even. So this is an
opportunity for us to organize and to present a coherent
explanation of our concerns with respect to the regulations that
are being proposed. So, with that, let me introduce our
fantastic panel that we have here. In alphabetical order, for
the folks that are here at the table first. Dave Aitel is an
offensive security expert whose company, immunity is hired by
major companies to try to hack their computer networks to find
and fix as a result vulnerabilities that hackers and
adversaries could use. The company is well-known for the
developing several advanced hacking tools used by the
security industry such as Swarm, canvass, Accomplice, Spike and
Unmask and Innuendo.>>Matt Blaze is a Professor at the
University of Pennsylvania and his research focuses on
architecture and design of secure systems based on crypt
graphic techniques and finding new crypto graphic techniques as
you surely mist know, he discovered a serious flaw in the
U.S. government’s cliprer encryption system. He is
interested in the use of encryption in various security
systems and controlled an attack against virtually all mechanical
blocks. Nate Cardozo is a Staff Attorney with the Electronic
Frontier Foundation. He focuses on the intersection of
technology privacy through expression. Defended the right
to blog and sued the U.S. government and lobby Congressed
for American surveillance flaws and in addition he works on the
E if. F Coder’s Rights Project counseling hackers academics and
security professionals at all stages of research. Mara Tam is
a researcher and historian of policy justice, culture and
security authored and coauthored and contributed research for
technical policy papers in arms control and security. After
earning a degree in art history, she works in bilateral
negotiations between the United States and India and has been a
panelist academic conferences and language and history
including the security initiative convened by NATO and
the European Science Foundation. And today we have with us also
by remote presence, very special guest, Catherine Wheeler of the
Bureau of Industry and Security. We are momentarily challenged
with the AV. We’ll see her later. But she is with us and we
can see her and she can hear us and she will be able to hear
you. So, I can tell you about her. She has served as the
Director of The information technology control division of
the Bureau of Industry and securities Office of National
security and technology transfer control since 2006. She is in
charge of this. This is great. She was detailed to serve as
acting Chair of the Operating Committee, reviews agencies. And
her experience here is incredibly relevant for this
task that we should here of trying to get to rules that our
community can live with and that also is also forward missions of
the BIS. So, we can’t thank her enough for taking the time to be
with us today. And I’m Moderator for this lovely panel, I’m Jim
Denaro, data security and intellectual property attorney
and advise hackers in how to stay out of trouble or get in
less trouble. So, with that said, let’s dive into the meat
of this. This is really about export control. And it’s a
really kind of in the weeds subject. And to really
understand the significance of the rulings that are being
proposed and how they would affect the community in such a
fundamental way, is worth taking a moment to explore what is
export control? How does it apply to you? How could it apply
to you and why does it matter? So a few notes of background
before we get into the meat of this particular situation here.
So, the U.S. government controls the exports of sensitive
equipment, software and technology as a means of
promoting national security interests and informed policy
objectives. This controls achieved by requiring people and
companies to apply for licensing before exporting articles that
are covered by the rules. So the question is, what is covered by
export control? That is really what the debate is here. So, at
a high level, we can break it into two categories. Favorite
additional defense articles that is not at issue here, but they
have their own licensing regime. Things that have no commercial
application. They are covered by the International traffic and
arms regulations. These include for example, armored combat
ground vehicles, tanks, as well as something more relevant for
us here, computers that are specifically designed or
developed for military applications. Second, our space
here, we have items that are considered to have both
commercial and military application. These are
considered to be dual use items and that is a term of art. And
they are controlled by the Export Administration
regulations for software and technology. Including things
such as high performance computers and encryption which
many of you have probably come across already. Infarcux the
schedule of controlled goods has a schedule titled, information
security. So you already kind of in range here. But those rules
are really about photography and that’s what you see the most.
So, for these dual use licenses, the U.S. Department of Commerce
receives somewhere between 12,000 to 14,000 applications a
year for this type of export activity. And compliance
matters. For these dual use export control violations,
criminal penalties can reach a maximum of $500,000 per
violation and an individual person can get up to 10 years in
prison. They can also be the subject of civil fines up to
12,000 dollars per violation and denial of export privileges. And
some cases both and I have criminal cases can be brought.
So — both and I have criminal cases — so the steaks are
fairly high at this point. So far, in the world we live in,
nobody in the community has been concerned that exploits or zero
days or even things like the hacking team system, were the
subject of much export control if any at all. So, there was no
concern there. But, that is changing. That brings us to
today and what you may have heard of as the Wassenaar
Arrangement. The Wassenaar Arrangement is 41 countries that
agreed to control certain dual use items. The U.S. participates
in this group and the list of controlled goods is updated
every year. Here is where it gets interesting N2013, the
Wassenaar Arrangement agreed to add certain things to the list.
And this is the text of the Wassenaar Arrangement. You
wouldn’t expect you to read this now and it’s too small anyway,
but the key here, the language is that intrusion software will
be regulated as a dual use item and as you can, I’m sure 1000
questions occur to you. What is intrusion software? And so on.
So that is really the item here that we are interested in. So,
the U.S. has committed to implementing Wassenaar
Arrangement agreement at the national level here in the U.S
so, rules will have to be written and enforced here for
all of us in the U.S. that regulate this. And on may 20,
2015, the VIS published the proposed rules for implementing
this Wassenaar Arrangement locally and most notably and of
particular concern to those in the information security and
research community, these rules seemed to go beyond the Barry
acquirements of this — bare requirements of this simple
statement here, and that is cause of much concern that we’ll
be addressing in particular here. So comments were taken on
the proposed rules and in light of those comments, some things
are happening and we’ll hear about what the response to those
comments has been. So that is the broad outline what have it
is we are talking about and why we are here today and why it
matters. So, here is the plan. We are going to go forward on
the panel, each panelist here comes from a different
perspective and has some brief opening remarks. So, we are
going to hear those remarks by various panelists and then we’ll
dive into some more questions and hopefully things will get
pretty engaging. So with that, Mara is going to kick it off>>
MARA TAM: So this is a quick and dirty introduction to dual use
export controls. What are they for? Basically to avoid this.
They are controls designed to monitor and regulate the
ecosystems around weapons of mass destruction or at least
that is where we get our modern export control regimes from.
Here is some of the stuff that export control is supposed to
regulate. Weapons of mass casualty, weapons of mass
destruction, which probably doesn’t mean what you think it
means. Disruption means something a lot worse than the
Uber definition. A puppy. So, the core logic of export control
is nonproliferation. It is controlling the spread of
dangerous technologies and this is done through a couple of
mechanisms. One of them is through knowledge, which is
deemed export this is going to be of concern to a lot of people
in this room and then the transfer of stuff or required
stuff. And this is where we get into dual use. And we identify
choke point technologies for this. We want to find something
where if you control it, you can control further progress in the
development cycle. These are difficult identify. They have to
be rare and conspicuous because you need to be able to control
every iteration of it or close to it. So you can see why for
intrusion and surveillance software, that principle sort of
falls a part immediately. Command and delivery platforms
just, they are too ubiquitous they don’t — that doesn’t work.
So here is a short history of the sort of dual-use we have to
work with and the International agreements for export control
that have happened sort of in the modern era. We started off
with the OEC, which this was an off shoot of marshal plan and,
they turned into the OECD. And their counterpart is the council
for mutually economic assistance, CME, because we are
adults, decided to call ComiCon. So arms exports to ComiCon
remember controlled under CoCom. These are the original CoCom
countries. So this is what he we had from about just after the
second World War until the Mid 90s. And successor to CoCOm is
the Wassenaar Arrangement which mashes together Comiconn and
OECD in this lovely mix and this is what we are stuck with now.
And these are all of the U.S. task force agencies tasked with
export control reforms like all great bureaucratic disasters,
this one was inherited from the Cold War and this is one of the
issues that we have right now is that there are so many people
involved in this process that getting good regulation is
really hard. So, the question I want to leave you with is, why
is a bug like a bomb? What is it about intrusion and surveillance
software and exploits that lends them or does not lend them to
regulations under a dual-use export control regime?>>With
that, we’ll switch to Randy Wheeler. So let’s see how we are
going to take this.>>Randy can you say something?>>RANDY
WHEELER: Yes, I can say something. Can everybody hear
me?>>She doesn’t look like a dog. She is a human being.>>
RANDY WHEELER: I can hear lots of people. (Applause) What just
happened?>>MARA TAM: We finally got through to you>>
RANDY WHEELER: Me personally or both?>>MARA TAM: You,
yourself.>>RANDY WHEELER: Oh, my goodness.>>MARA TAM: I
realize you can’t see everybody but wave hello to DEF CON.>>
RANDY WHEELER: Hello DEF CON 23.>>Don’t turn it around. Bad
idea. (Laughs)>>So, technical issues, we cannot see randy and
her slides at the same time. So now that we all had a chance to
or at least you had a chance to say hi to her, we are going to
have to switch over to her slides so you’ll hear her.
You’ll hear her but not see her.>>This is a two-hour meeting
for obvious reasons ( Laughs )>>RANDY WHEELER: All set? Thank
you very much for inviting me to participate in this panel. I
really appreciate the opportunity to address folks at
DEF CON. And I’m going to give a very, very brief overview of the
proposed controls on intrusion software items and IP network
systems and the Export Administration regulations that
Nate mentioned earlier. Next slide. So, my next slide isn’t
working. So as Nate mention the in the Export Administration
regulations, we have national security controls on computers,
telecommunications and information security. These
listed items appear in the commerce control list, which is
part of the Export Administration regulations. And
there are other categories as well. The category 4 by part one
and two controls are a responsibility in my division,
information technology controls division. We process
approximately 2500 export license applications and also
2000 commodity classification requests per year. To date, most
of our work has been in the encryption area in the category
5 Part II. Partly because over the past several years, as
everybody knows, everything has encryption in it and so, items
that would have been in the category 4 or category 5 part
one, have moved over into the encryption control section.
Within each category, as Nate mentioned, for commodity, test
equipment and software and technology. The information
technology controls division comprises 9 licensing officers,
including myself, and we have three electronics engineers on
the staff and six export policy analysts. Next slide, please.
The new control entries, the subject of the proposed rule,
are three related list entries in category 4, Nate said system,
equipment, software, components, especially designed modified for
the generation or operation of or communication with intrusion
software. We also have a separate technology control or
technology required by the development of said intrusion
software. And then, the proposed rule also includes a definition
of intrusion software. There is also a separate entry for the
network communications surveillance system in category
5 part one, telecommunications. Next slide, please. As Nate
noted, the control list entries were proposed in the Wassenaar
Arrangement in 200013, and they were adopted by the Plenary in
December, 2013. It is worth noting that the category 4 and
category 5 proposals were submitted by two different
countries aimed at covering two different types of products. And
the interesting thing about them that was they both had an
element of human rights in the purpose of the control, the
category 4 controls were aimed at offensive systems that are
being sold not on the commercial market but directly to
governments, potentially repressive regimes to be used
against their citizens. And the same element was present in the
proposal for the category 5 part 1 monitoring surveillance
systems. Once the Wassenaar Arrangement agrees to a new
Creole list entry, it’s added to the multi-lateral Wassenaar
control list. And then it is up to each member country to
implement the control in its own list pursuant to its own
statutory and regulatory authorities. In the United
States, the dual use list for national security product is
implemented in the commerce control list so the process is
to draft a rule and to issue the rule, usually as a final rule
usually in the may or June timeframe in a year following
the adoption on the Wassenaar list. Between December 2013 and
may 2015, there was a great deal of interagency discussion on how
to implement these new control list entrees. In the
administration regulations, we have a reason for control for
several reasons for control for the same item. We need to
determine the licensing policy, license exceptions that may
apply, and in this case, we needed to consider that there
was overlap with existing encryption controls. As I
mentioned earlier, a lot of products moved over into
category 5 Part II over the past years because they have added
encryption N this cases, we already had controls on
penetration testing products that included encryption and at
times cryptanalytic functionality, and have been
licensing them under category 5 part 2. So part of the question
was, what do we do with those products? Do we change the
treatment of them? And in the proposed rule, there is a much
tighter restriction on the export of all products that
could be described under the new control list entrees, including
the penetration testing products. We published the
proposed rule in may 2015 with the request for public comments.
And boy did we receive comments. We had received almost 300
comments totaling some almost 1000 pages. Many of them were
very thoughtful. Before the comment period was over, we
received many requests to meet with various groups and industry
coalitions and so forth. We were very, very grateful that there
was such interest in talking to us and explaining the issues
that the proposed rule raises. There are three areas that the
comments have greatest. The first was the implementation in
the proposed rule and as I mentioned, the restrictive
license requirements and no availability of license
exceptions which places expert license requirement on all
destinations except Canada, and all government and
non-government end users and would require an export license
for intra-company and internal use in companies for technology
and software, and it would also impose a license requirement on
deemed exports, as Nate mentioned very briefly. The
release of technology or source code, to a foreign national in
the United States is considered to be an extra to the home
country of the foreign national, and we do receive a fairly large
number of deemed expert license applications each year by
companies who want to release technology to employees who are
not U.S. nationals. And these export license requirements
would apply to the new control list entrees without any
exception. The proposed rule also is set forth a very
restrictive licensing policy with approval only to poor
countries and case-by-case to all other destinations, in
addition to the national security reason for control, it
imposed a regional stability reason for control, which is
very restrictive and it set forth a licensing policy under
the regional stability provisions of regulations.
Finally, the proposed rule set forth a denial policy for
products with zero day or route kit functionality. These terms
did not appear in the Wassenaar text. This is in addition on a
licensing policy basis in the proposed rule. Second, and we
were expecting the comments on the restrictive proposed
implementation, but we also received a very large set of
comments on the text of the Wassenaar control list entries
as well, in particular the definition of intrusion software
raises many questions and issues and the other panelists will
address some of those and their many concerns about the scope of
control on technology for the development of intrusion
software as defined. Finally, there were other issues raised
even beyond the Wassenaar text that are very important to
consider. The likelihood that the imposition of these controls
would achieve the purpose of addressing Human Rights and the
likelihood that they would even cause more harm to security
research generally. In addition, there are a number of comments
that noted that the restriction on sharing of technology on
cybersecurity research appears to be at cross purposes with
other government initiatives, including pending legislation to
encourage the sharing of such information. I forgot to tell
you to change the slide. I’m sorry. So, we are now at the
very last slide that says, next steps. The next steps in the
regulatory process, we are in the process of reviewing the
comments and again, we do appreciate all the time and
effort that all types of companies and researchers and
industry representatives and industry coalitions took to put
their thoughts down on paper. We are planning to discuss the
comments, the issues raised in the comments in a series of
technical Advisory Committee meetings in the rest of the
calendar year, and although Mara mentioned that there are so many
government agencies involved in export control, we found that in
this process, there were a number of government agencies
who are working with expertise in the cybersecurity area who
were not involved in the development of the rule. And we
hope to have them participate with us in the open discussions
with the constituencies who are interested in the issue in the
open meetings in the text call advisory committees for the rest
of the calendar year. Also, given the issues raised, we will
consult with our Wassenaar partners a number of the other
member countries have already implemented these control list
entries from their national control lists and apparently,
without some of the reaction that we have received when we
published the proposed rule. So, we would like to talk to them
about the entrees and find out how the implementation is
affecting their industries and research communities as well.
Following these three steps, we intend to draft a revised
proposed rule and again we would have opportunities for public
comments before we would publish a final rule that would go into
affect. Thank you for inviting me to participate and I look
forward to hearing the other panel member’s presentations.
(Applause)>>Thank you for that. That was a helpful
explanation and thank you to the members of the audience also for
staying with us for this explanation of what if is we are
talking about and what the rules are and how this process moves
forward this is a back and forth process between research
community and many other stakeholders that are interested
in how the technologies that are used in surveillance software
may be regulated on a global scale. So this is the framework.
These are the parameters that we are working with and with that,
we can take a deeper dive into how the proposed rules are going
to potentially have some very significant impacts on the
various interests. So with that, Nate, take it away>>NATE
CARDOZO: Sure. Thank you. I’m a Staff Attorney with the
Electronic Frontier Foundation. As Jim mentioned earlier, I love
technology so I’m going pull up my notes a phone and do the
slides from the computer. Because we can’t do both. John
Gilmore in 1993 or there about, told us that the net interpret
censorship and damaging routes around it. That statement is as
true today as it was more than 20 years ago when Gilmore told
us and it is far more true today than it was then. Back in the
90s, the export of — and this is a gross over simple
indication of the expert of cryptography was controlled
under itar, under the United States munitions list. As a
weapon. So this slide could not be exported from the United
States. Nowadays, we are left with the Wassenaar Arrangement.
EFF sued on behalf of Dan Bernstein in the 90s. We won. We
got a ruling that said, code is speech. And cryptography was
moved out of I tar and into the EAR. Export Administration
regulations. Now of course we are dealing with Wassenaar. Why?
This is the problem that Wassenaar was designed to solve.
That is enigma machine designed to protect German banking. It
was a commercial encryption device that was of course
repurposed during the war at first to great affect. This is
also the problem that Wassenaar was designed to solve. Not
really of course. The maker bot is not controlled under
Wassenaar. But guns are. Not guns, per se but nerve gas and
precursors, et cetera. But what about information? How do you
control the export of information? And I would propose
that it’s not going to work any better this time than it did the
last time. Because we have things like this. I can export
information very, very easily. But what do we do? There is an
actual problem smear it’s a significant one. What do woo do
about things like this? Packing team, these are pieces of
software that I really don’t want in the hands of repressive
regimes around the world. What do we do about it? As Randy
said, one of the things about the way that export controls
work especially in the United States and the way the proposed
rule that we are talking about today works, is that it controls
exports period to anyone talking to your coworker who is not a
U.S. person, that is controlled it doesn’t matter if you’re
selling thin fisher to the government in Ethiopia or
selling medicine to a pin tester in Chile. Those are both
controlled. One of those uses I’m fine with. The other one,
I’m not so happy about. But there are already tools
available I would suggest that going to end use or end user
control is a lot better, right? This is an actual Cisco slide
talking about how Cisco is going to help the Chinese government
build a golden firewall to combat evil religion and other
hostilities. This kind of thing is what we should be worried
about. We should be worried about our technology companies
building the tools as Human Rights abuse. The Wassenaar
Arrangement is intended to control things like this but it
ends up sweeping way too much. Because it doesn’t take an end
user control. Here is another thing that I’m worried about
this is a hacking team e-mail talking about sales to the
government in Ethiopia. I feel – Electronic Frontier Foundation
and representing Ethiopian suing the government for wiretapping
his Skype calls. So I would propose to you that there are
other tools besides a blanket export control regime that are
better suited to holding companies responsible for doing
things like building the great firewall of China in the
specific evil religion plug in that Cisco built or thin fisher
selling to the government of Ethiopia with full knowledge
that it was being used against journalists activists,
dissidents and — so that is where I come from. And I’ll turn
it over to — who goes next? Matt? David? (Applause)>>
DAVID: I’m going to start off real quick with, I guess, a bio,
in case you forgot who I am and why I’m here talking to you
guys. And the reason for that is that my first employer out of —
well, during college, was the National Security Agency. And I
since started immunity, which is a company many of you know of
because we have a free debugger, which is surprising to me but
that shows how awesome my marketing skills are. I also
have a mailing list called daily Dave, which is discussing a lot
of this Wassenaar. I can’t pronounce it properly. Activity.
And we became very concerned when we first saw it coming down
the pike. In particular, because we sell to the general public 3
or 4 major tools. We have canvass, which competes with —
and core impact and I assume many of you have used one of
these tools to do operational penetration testing, which is
something that is required by PCI or required by HIPPA or
almost everything that is security related. Of course we
also sell Silica, which does wireless penetration testing
which qualifies as a crypto analytic tool under the BIS
regulations. We also have a conference called infiltrate
which focuses on offensive and attack technologies and offers
people a way to be very honest about what it is we do. And so,
my whole life has been spent building command and delivery
platforms essentially and that is the exact sort of behavior
that these people, some people find uncomfortable which is a
necessary part of our existence in order to understand and
secure ourselves. It’s been said that pressie won’t come up on
his laptop but also been said that defense is the child of
offense and so, for those of us in this room who work on
offensive things, I think we can all spend one hour of our time
to reply to the simple to use website and it surprised me more
than anybody, that BIS has an amazingly easy to use website
for submitting comments. You can read the regulation about 15
minutes. You’ll never understand it so don’t even try. But you
can read it and then you can write comments on it that say
what or how it would affect your daily life and it will take but
an hour. You can do it during Simpons reruns or something. So
make it funny. Just don’t include curse words or anything
crazy. And I think the next round for comments should not be
1000 pages. I think it should be 100,000 pages. I think that
Randy would very much enjoy having everyone at this
conference, everyone here is impacted by this rule in a major
way. The only reason I’m involved is because we pay our
lawyers a lot of money to keep us out of trouble. But no one in
this room wants to pay these lawyers all that mon– they do.
The lawyers do. I’m not a lawyer. But the lawyers would
enjoy that. And I don’t think you should have to. And I think
it’s a uniquely-unAmerican thing to control the export of
information, which in a sense, the human voice is the original
export technology for information, and I think we
should try to keep that voice free from any kind of
overbearing regulation as a matter of course. We almost have
pressie. It’s amazing. I can go on for hours. So here is my per
suspect. And it’s your perspective at the end of my 5
minutes. Which is that export control is a bad idea for
anything in this area and we are talking a lot about the
intrusion software part of it. Let me say it is already trying
to frame the discussion because when they say intrusion
software, they mean anything that does anything useful in
security. When they say surveillance software, they also
link in anything that does intrusion detection and
anticrime wear on any scale. And I’m going to talk more about
that. But this, I believe, it should be and is difficulty —
Randy can’t see it. I’m sorry. It says here, you can see it?
No? She seen it already. Okay. So Thomas Jefferson, among many
things should be our guiding light when it comes to
protecting ourselves against attorney and we should avoid
ourselves becoming the form of tyranny. And that is what they
are doing us to do. And if you read the definitions in the
thing, it should scare you not that the definitions are there,
but that they were ever allowed to be put into the regulation at
all. Something went horribly wrong with the whole process.
And I’m going give you an example that no one talked about
yet, which is carrier grade. Forever those who have ever
worked in telecommunications, which is a lot of you, carrier
grade by definition means reliable. It’s a marketing term.
And how I think it got in the regulations is that I think
privacy International used it in a random report. They are like,
we are scared of anything carrier grade. But carrier grade
is not a metric for speed, yet if you — if I made you zoom in
on this thing, in the actual defense of the regulation that
BIS had, they said, well, we think it’s anything fast enough
for a city or a country but we won’t put an actual number on
it. And the reason for that is because there is no number. And
if you did put a number on it, it would have to go up
exponentially over time. I love in south beach. Not that I’m
recruiting because my company awesome. But you shouldn’t move
to it for south beach. But south beach has like every apartment
can get 500 megabytes your door via a mesh network someone set
up. You can do the same thing in New York and San Francisco. And
at what speed is carrier class? We are a small city. So, I don’t
understand what the bar is. There is no bar what they mean
is, we mean what we mean when we say what we mean. Right? And
that is, this should scare you because the penalties are so
high for all of us, for breaking these regulations, that you’re
guaranteed to break them and you’re guaranteed to be under
that own us. What is a route kit? The it’s not in there. This
was a program — if this was a program, this document would
have never compiled. Support zero to exploitation. First of
all, zero day is not a term you can define because it means
something you don’t know. And everyone has different amounts
of knowledge so things that one of you knows may not be a zero
day to me. They may be something I have sitting around that I
don’t think is important. And so to support zero day, simply
means you can run a program. So everything that qualifies as a
command and delivery platform can in fact be modular and run
programs this is an extremely low bar and yet it is under the
default denial section of the regulation, which means that at
some point, they thought this will be fine. And that is just
the beginning. Here is what is going to happen with the next
regulation they come out with. There will be a million more
examples just like this. We have a process that is creating
programs that cannot compile and making them with laws with hew
ming us penalties. That’s what is broken here. And the
overreach in this area has massive, massive dangerous
implications. Deemed exports alone means those who have H1Bs
are cast out of our community as per eyas. Technical data is
something that you, as a human being, cannot understand but the
lawyers among us will argue about for years at 1000 dollars
an hour to tell you if you’re allowed to open your mouth and
talk not person next to you. Required for — again, some of
these phrases should scare you because if you as a person,
can’t understand if what you’re creating and exporting is
required for the building and delivery of command and delivery
systems, then you’re at risk no matter what you do. And that is
what this regulation does. It puts all of us under this giant
sword so the people who knock on your door, can say by the way, I
noticed you were violating the law. We love to you cooperate on
something else. That would be awesome. I can make this stuff
go away. And there was a very bizarre section in the
regulation when they went to defend it on their phone calls
as they started getting some heat, which said that if you
release it to the public or vendor you’re okay. But if you
release it to just private industry, you’re not okay. And
we are talking about some value decisions in the disclosure
arguments that don’t reflect this community at all and don’t
reflect the industry at all. And again, just to nail this point
down, penetration testing software, which is this current
regulation would have been restricted as much as a nuclear
bomb, is a required operational practice for every company in
America. And I think we talked briefly, especially Mara did,
that export control if you’re going to apply it, should at
least have some hope of accomplishing the desired goals.
I don’t believe that the desired goals are worth accomplishing
but I want to run this down here. Here is how you protect
those poor journalists and activist against thin fisher and
gamma. Give them an iPad. Because neither can attack
unpatched iPads. So that is cheap. I’m willing to donate
iPads to these people to avoid regulation because I think it is
a cheap way to do it. Here is what you don’t do. Ban all
software that makes you uncomfortable at great costs to
the rest of the world. And I think we should talk a little
bit about licensing because even permissive licensing kills sales
and retards innovation. Because in order to go through the
encryption controls, you currently have to wait one month
after developing your software and this is almost all software
because the rule is, anything that links the live SSV under
this rule. And if you do anything to your crip tow that
changes your crypto or how to use your crip tow, you’re
supposed to send them a note and explain it and describe it and
wait 30 days and then you can do a release. And so if you
wondered why core impact and canvass are on a monthly release
cycle, this is why. And it’s extremely difficult to innovate
under this kind of condition. And of course, anyone actually
malicious f there was a malicious Ethiopian person that
Nate doesn’t like for some reason, then they could always
get a account. That’s what they are going to do. Even at the
best chances, there is no way export control could work even
if it was meant to work, which it is not. So, I think this
community, all 700 people in here, are largely of the opinion
that code is not a weapon. Code is speech. And I think part of
the reason of that is, we understand something at a much
more basic level, which is that you can break down any fact into
an infinite number of smaller facts, which you can then
combine in combinations to produce the original fact. So
for example, if I was going to write a paper on if you have the
extended instruction pointer, then you can use a certain
technique to by pass ASLR and then I would write a separate
paper on, here is how I would get the IP using adobe reader
and a particular technique. And if I could bind those things up,
those are controllable. But if you I don’t, they are not
controllable. That’s the key problem with regulation in any
space where we are trying to regulate speech in this way. And
of course, the irony in this is that when you see people who are
privacy activists, a spousing these kinds of controls, they
are not looking forward to the obvious next step, which is to
enforce them, you need a global surveillance network, which is a
horrible thing to have to put in to their hats. So in summary,
their idea is bad and they should feel bad. And in the end,
what is going do happen if this stuff goes through as is, or
even close to as is, is it that all of you are going to feel
bad. So I’m hoping that everyone takes that hour to comment on
the next one and we can further influence it by means of killing
it. And that’s what I have got. And hopefully everyone agrees
with me and we can all go. (Applause)>>So, I’m Matt and I
should say in spite of the introduction, I’m not a lawyer.
Though I do occasionally impersonate one. I’m a Computer
Science Professor. And one question is, what am I doing
here? I am working in this abstract field and I’m not
directly a target of these regulations in the sense that
nobody thinks that what I do and what people like me do is bad
and needs to be regulated. I mean, the worst people say about
what I do is that it is useless and stupid. But I don’t think
anybody says that what I do is harmful. And I don’t think even
the Wassenaar advocates think that academic published research
in this area is something that is supposed to be regulated or
at least that’s not a particularly common feeling. So,
it would be very easy for me as an academic to say this is
something that I should sit out and watch and let people with
the vested interest like Dave, fight this out for their
interests. And in particular, the work that I do, when you
look a little closer at how it actually gets done and how these
regulations to be implemented, particularly over time I start
to become a lot more worried. And one reason is that, my job
is to think of things and publish papers for the greater
good and I publish things and fundamentally that’s a defensive
activity. The more we learn about what to do, its more
robust systems we can build. But at the level of work that we are
doing, the distinction is meaningless. We can’t study
defense without studying offense and in fact if you look at the
papers that we publish, we tend to flip around between defense
work and overtly defense, overtly offense back and forth.
Somebody publishes defense, and then attack and at the end of
that arms race we end up with something a little bit stronger.
So fundamentally, I’m in the offense business as much as I’m
in the defense business. Another thing that should reassure me is
I don’t produce products or export things or sell things.
But it is true that fundamentally what we are doing
is not producing in the academic research world. We are not
producing code that we are selling to people or code we are
incorporating into attack products. But when you look at
the process, there is quite a bit of code exchanged and there
is quite a bit of exporting going on. About half depending
on your institution, will go up or down but it’s certainly in
the ballpark. About half of our graduate students are foreign
nationals and that’s generally true at any research oriented
University. People come to the United States to study this
stuff. We have colleagues in another countries that we
collaborate with and the process of producing research is often
involved with a process of experimentation exchanging code
and working on things. The export regulations, effectively
limit what I can say privately with my colleagues prior to
publication, and that means essentially it’s not regulating
the output of my work, it’s regulating the process of doing
my work in order to produce that output. So people who say, you
don’t have to worry because you’re papers are published by
the first amendment, you don’t have to worry because this only
affects attack tools and you’re not selling attack tools. And
you’re not exporting things over — that’s true about the output
but not true about the process necessarily. So even though
there are many reassuring reasons to think that this is
work that shouldn’t or that I and people like me shouldn’t
worry about, when we drill down to the actuality process, this
is something that me and my colleagues have to be worried
about every day. Now I’m lucky that I work for a big fancy
pants institution that can afford lawyers. And fortunately,
at my institution, the lawyers that we employ generally, see as
their job finding ways for me to do my work instead of finding
ways to stop me from doing work. And as soon as — but as soon as
I talk to them about export rules, that flips. The answer
tends to be, you’re taking some risk here. You need to worry
about that. We better go and get a license to do this before you
do that. Unfortunately, I have the support where they will help
me with this, but these are extremely difficult rules to
comply with even in the easy case where you know that you
don’t have to make an argument, where you just have to go
through the motion. Many people who are doing research of the —
at the same caliber or higher than people like me at
universities, aren’t affiliated with universities and don’t have
that kind of institutional support. So for me, with
institutional support, it’s hard. Somebody without
institutional support, it becomes kind of a death nail.
Now the last thing I worry about is, as a Veteran of cryptowars 1
in the 1990s before we needed to number the cryptowars, the
primary thing we were talking about was export law.
Cryptography was covered under ITAR and the lever the
government had to regulate cryptography was not that there
were rules about using cryptography domestically, but
that there were rules about using cryptography
Internationally. And that was what we were talking about in
the first cryptowars. Now we won that and now we — largely
deregulated most consumer-grade and research-grade crypt oh, but
what that illustrates to me is the way that regulations that
are intended to accomplish one set of policy goals here when
they are implemented, in the future, can be used to
accomplish other policy goals that weren’t even on the table
or being considered by the people proposing them. And I
worry here, that today we look at this and we say, nobody is
meaning to regulate academic inquiry into computer security.
That may not be true 10 years from now under the trump
administration or what have you. And, these rules may change in
the regulatory tone may change later. So, this is something
that I find worth engaging in and I think you need to consider
whether it is something that you need to engage in as well. So
thanks. (Applause)>>So I also note that we have these little
buzzers that make funny noises and we were supposed to press
them if anybody disagrees with each other. And nobody seems to
disagree with anything any of us said.>>I disagree we won the
cryptowar. I think we actually lost. So two of you said we won.
But when you sell software, anywhere, in the country
externally, bays every piece of software uses crypto in some
way, you’re under very strict regulatory frameworks and as
much as like you’re going to get a license, the fact is, you’re
sales process is going to be pretty messed up. You’re sending
away to the government a list of all of your customers, which
some of you may feel uncomfortable with and many
other regulatory issues with even understanding. These are
not simple. These are some of the most complex convoluted laws
on the planet that you, as a simple researcher, are now being
required to understand or else be under severe penalty. The
same thing true of kepto. I think we lost. That’s my
personal opinion>>Let me jump on something that Dave just
said. The rules are very difficult to understand. And I’m
a lawyer so I’m going to look at this through a U.S.
constitutional law perspective. And this is again, going to be a
gross over simple indication. In constitutional law in the U.S.,
we have a doctrine called, void for vagueness. If a criminal law
is vague enough that an average person of ordinary intelligence
can’t tell whether their conduct would be criminalized or not,
that law fails constitutional scrutiny. We have seen that it
is most common in hate speech or in excitement context. But it
works here too. If an ordinary person of average intelligence
reads the Wassenaar control list and can’t understand them, then
the implementation of those control list would be denial of
due process and unconstitutional>>And that gets to one of the
sort of core issues about export control which is like I said
earlier, you can’t control something if it is — you can’t
choose a choke point technology if it is ubiquitous. So when
something is omnipresent liken corruption like the command and
delivery platforms, you run into the same problem. You don’t
know. And therefore the control fails.>>I think it is telling
that in fact, BIS has on their website web applications that
run you through an expert system to determine if certain phrases
apply to you. Such as required for, or as needed by. There are
little phrases in the regulation that you cannot understand. Only
the expert system can understand. And I think they are
minute to help you but they design — meant to help you but
they demonstrate the design of the arguments is already vague
and if you talk to your local export control individual, which
unfortunately immunity gets the privilege of doing a lot, they
will tell you as well that even the lawyers underbuys don’t
really have a clear understanding of it. That they
can explain to you, for example, what soft swear meant to be old
and what is not. Because these issues are so complex and they
are rarely going to court it’s been really rare to see the
crypto stuff result in a penalty against a company. But that’s
not as important as whether or not it is used as a hammer in
general, which I think should scare you more.>>It’s pretty
well established that the rules are intended to prevent the
availability of surveillance software to repressive regimes.
But there are questions about whether or not these rules are
effective in doing that and whether they would also sweep in
lots of legitimate software at the same time, if we could use
that term. So, I’d like to give Randy an opportunity to respond
to that and sort of give more context into how the rules are
being tailored to cover just what the original intent was.>>
RANDY WHEELER: I think the comments are right on point,
that the Wassenaar control list attempts to describe particular
products, particular functionalities, and the in
stent to narrowly define what was going to be controlled, but
in fact, what we have learned from the public comment process,
is that either the language is not well stated so that
reasonable people with potentially different
vocabularies are reading the language differently, and as
well a number of unknown or unexpected products or
activities are being swept into the control and that’s we want
to address going forward. Is there a way to capture only the
products that we are interested in captureing and only licensing
those exports that are of concern. Certainly from an
administrative law perspective, and as a regulator, I think it’s
poor use of government resources and a very poor use of company
and industry and researcher resources to — (Broken audio)
>>She is gone. We lost her. Network resources, too.>>Who
wants to say something controversial?>>I think she
might be back. Hold on. Are you back?>>RANDY WHEELER: I’m back
>>You said resources and then you disappeared>>RANDY
WHEELER: Sorry. I just meant that I think it is a poor use of
everybody’s resources, both the government resources and
industry or researchers resources to spend time worrying
about transactions, export transactions, or deemed export
transactions, that are subject to a policy of approval. There
is no point in requiring licenses for those types of
activities and so we should work to only cover those transaction
that is would be of concern>>So in order to cover just those
certain transactions, it seems like it is a project of
definitions. And, a lot of what the concern is, how intrusion
software is being defined and I think there is a bigger question
as to whether or not intrusion software is — of any kind of
meaningful definition. So, I’d like to open that up to Randy
first, and also I’d like to hear from the rest of the panel about
if there is anything to be had there.>>RANDY WHEELER: I would
quickly agree with you from the comments we received. It is
problematic definition. Again, the people who are in — we have
government regulators trying to define this and then when people
who actually deal in the products and technology that
look up the definition, it either doesn’t — they don’t
understand it what it was intended to do or they used the
vocabulary differently and that is up for regulation then if
there is lack of understanding of what it covers. And
particularly if it is understood to be broader than it was
supposed to be, then it needs to be revised. The frequently asked
questions were an attempt to address that but we got to the
point where even in the answers to the questions, that we posted
our website, we were referring back to the regulatory language,
and we just kind of got stuck because we didn’t have the
correct vocab layer tow address the issues that were being
raised. So that is what we hope to look into in the next step of
the discussions. Thank you.>>So one of the things that we
asked in our comments to BIS, which was also echoed by Google
among others, is that congress department and I guess, the
State Department, go back to the Wassenaar Arrangement itself.
The next meeting is at the end of this year and work on
clarifying Americans through BIS but working with the 41 member
states of the Wassenaar Arrangements to add clarity to
the control list there. Software that modifies the standard
execution path of a program. What does that mean? Why are we
focusing on that? And that is not something that BIS can do
alone. That is something that needs to go back to the
Wassenaar Arrangement itself. So that is our best case scenario
if BIS didn’t just do a revised proposed rule and open it back
up for comments but that BIS and State Department go to Wassenaar
and change the control list there to make it better and then
do revised proposed ruling>>And by make them better, he
means let’s just remove this. Because there is no good way to
do this. What you hear from people — he doesn’t agree but
he’s wrong. If I agreed with him we would both be wrong and that
would be terrible. And here is the thing. They would say
regulation in this space is inevitable so you might as well
as an industry feel free to come up with language you’re willing
to be bound by. And I will tell you this, that is a fool’s
errand. And it is a trap you should not fall into and I think
even if you could describe all of today’s software you found,
the reality is, you’re also describing software that in the
next generation is going to be required for normal operational
business. Because is this say community that moves far faster
than regulation and always will and always should if we are
going to survive. And I think that when they say please
describe some language that works for us today, you should
say, I need language that works for us forever and it’s not
possible and therefore we should not do it.>>MARA TAM: Also
worth noting the convolution actioner rising from the
Wassenaar language is due in large part to the fact that
Wassenaar was never designed for Human Rights purposes. I mean,
this was a – auto export control regime that Wassenaar inherited
was all about controlling arms. And several advocacy groups,
namely International end cause, decision E. decision to get
these category 4 and 5 entries added and they were successful.
One of the irritating things about that is they knew that
Wassenaar was not fit for purchases. They knew that export
control would not work for these items. But they persisted. And
unfortunately, we are dealing with that right now and goods
intentions and all of that but this was not the right way to go
about it.>>From my perspective, it’s not the
software that is a problem. What hacking team does, what thin
fishy do, it’s a standard remote administration tool. You can use
any of the remote administration tools would have worked just as
well to spy on my client in the Ethiopia lawsuit. What we care
about, what matters isn’t the tool itself it’s the service
support and most importantly training that comes along with
it. Thin fisher doesn’t cost very much but getting your
intelligence agency all strained up to use it and the ongoing
support contract is what gamma makes its money on. That’s the
problem. These tools — it’s not the tool. It’s what goes — it’s
the infrastructure surrounding it. The Wassenaar Arrangement
was — sort of designed to take that into account. Intrusion
software is not controlled under the Wassenaar Arrangement it’s
the infra structure around intrusion software that is
controlled. Technology required for this et cetera. But without
tailoring it specifically to state uses. And it is those
state uses that we see causing significant harm out there in
the real world.>>Keep in mind, under U.S. law f I’m correct,
anything that is designed specifically for U.S. government
or military use would be controlled under ITAR and the
same — this is something that no one mentioned is that
actually half of the team was perfectly well regulated under
the Wassenaar and they went to the government and said can I
have a license? The government says yes you can for anyone you
want. So even if under the most strict interpretation of thieves
regulations, the reality is those chomps operated out of
smaller countries, which would be every company in this
business if U.S. decides to implement these regulations, can
easily go to their government and ask for an out anyway. So
even if there was a perfect language that applied only tow
really bad things, which we don’t know what are, but if
there was perfect language, it still wouldn’t work because you
would have every company going to their government saying, I
want an out>>An alternative you have to worry about pushing
these governments into capabilities developments and I
think Nate raises a good point which is it is the back end
support which leads these technologies to be so harmful in
those context. But if these states surveillance agencies are
no longer to buy off the rack, they will move to capabilities
development for themselves and that is a very serious problem.
There is no unwritten law of cyber cyb that says Bahrainian
engineers couldn’t come up with an equivalent of packing teams
RCS especially now the source code is leaked. So controlling
this from the Tom down simply will not work.>>Especially
when we are talking about activities that are done by 10
people with computers you can buy off the shelf. I think that
is — the inefficiency of regulation in this space can’t
be overstated.>>So we are getting a good sense of what the
objectives are. It would be great if you could fill us in on
where these objectives come from. I think a lot of people
might make the criticism that it may be — or ask the question as
to whether or not it is properly within the scope of the mission
of BIS or commerce or the government to be taking the
position as to what types of software should be made
available to any particular regime. So the question is,
where does these regulations of course don’t say on their face,
you can’t sell to a particular repressive regime and doesn’t
define who they are. It defines the thing. So if you can give us
insight into where the input on these particular sets of
regulations are coming from within the U.S>>>>So, there
isn’t an export control of community involved with the
Export Administration regulation prescribed by statute. State
Department, defense department, and Department of Commerce. And
we all provide expert group members to attend the Wassenaar
discussions. The consensuses? 2013, that — the consensus was
in 2013 there was a set of products that was of concern
within the scope of the Wassenaar man date, that
addresses dual use products that can be used by the military or
by civilian agencies for civilian uses and so that is how
the language was added to the Wassenaar list. And then I think
that it is fair to say that immediately, even though we have
the understanding at the time, what the products were described
in the language that that was not perhaps a good understanding
and the public comments certainly born that out that
there are many products in this space that could be considered
to be described in the language that were not intended to be
controlled under the controllist entrees. So we don’t have a
disagreement here. There was an intent to control certain
products but a good number of products were then swept in the
technical description and that is what we are dealing with now.
And all of the comments so far have echoed the comments we
received in the public process and we’ll be certainly taken
seriously under consideration going forward.>>So speaking of
— thank you for that answer. So, speaking of the public
process, we’d like to open up the floor here just to audience
questions. So they’ll have to — we don’t have a mic for the
audience so we’ll have to make do and repeat the questions and
we would love to hear your input and of course panelists, feel
free to jump in. Line up behind Joe.>>AUDIENCE MEMBER: (Off
mic)>>So the question was, is the only reason we control crypt
oh, the Wassenaar Arrangement? And then the second part is, is
there any good reason to control the export of crypt analysis? So
the answer to the first question is, no. We have controlled
crypto since — that’s a pre-Wassenaar thing. And then
the second half of it is, why do we still?>>MARA TAM: So
cryptography was controlled under CoCom, the predecessor to
the Wassenaar arrangement and it’s worth noting when
encryption came first came under export control, it was not as
sort of insane as it sounds now. I mean, encryption was a big boy
toy. It was something that nation states did. It was not —
in the era before personal computing it was not ubiquitous.
So, export control might have made sense at some point. I
don’t think it still does.>>And that was another thing which
I, in the comment section after the Electronic Frontier
Foundation said, is before we attempted to do anything more in
surveillance software, let’s decontrol encryption entirely.
I’m not sure they will do that but that was what I asked for>>
And crypto export controls are perfect examples of one policy
goal when the regulations were originally enacted to keep
crypto boxes out of the hands of military adversaries. Perfectly
good public policy goal if there are crypto boxes and military
adversaries that might be able to exploit them. And then
software got invented. And suddenly, we are now worrying
about law enforcement domestically and these
regulations that were enacted for a purpose completely
different from what they are being enforced for>>So Randy,
just quickly on the crypto subject, that is obviously not
part of Wassenaar, crypto has been regulated more tightly in
the past and the regulations we have now are relatively now more
relaxed. Can you give us insight into any trends that BIS with
respect to how crypto might be regulated going forward>>RANDY
WHEELER: Certainly a lot of changes to the encryption
entrees T is a Wassenaar control under category 5 part two. I
have been involved in the program, unfortunately we had a
series of de-controls in the encryption provisions. But in
the same way that we have the technical description issues in
the proposed control list entrees, we have them in the
encryption provisions as well. For example, I would point to a
couple of new decontrol notes, L and M, that we just implemented
in the regulations this May and again, they are technical
descriptions that are not exactly product descriptions and
we are in my office still trying to work through exactly what
product these de-control notes cover and don’t cover. And these
are decontrol notes L and M so that means there are several
others starting with A. And we go through all of this and it’s
a very broad control with many different carve outs and notes
and so forth. We have limited the encryption controls to
products whose primary function is communications computing
networking or information security, which makes
refrigerators not subject, that have the alarm system that is
have encryption. And that is a good thing that didn’t happen
until 2010. We are still working on that. We still would like to
have a positive list. We would welcome public participation in
that process as well to try to make the rules more concise, and
more understandable. There are many permissive provisions in
the encryption area. Many license exceptions that is very
broad and for example, it applies to almost all deemed
exports of technology. So, we have a very permissive regime in
the end but a lot of text to get there. And certainly it could
use a lot of improvement. I could talk about the encryption
controls all day. I have a day-long seminar that goes from
soup to nuts. And we would like to continue to improve them and
again, we welcome public participation through the
Advisory Committee process for that purpose>>Perhaps one day
there will be day-long seminars what intrusion software is. So,
we have got — we have a line up of questions. So we should take
the next one.>>AUDIENCE MEMBER: (Off mic)>>I think the
question is — sorry if I’m paraphrasing. I didn’t hear the
whole thing. I think you’re getting as, technology changed
and the sues of technology changed, are the regulations
still relevant or are the regulations following the
technology in an appropriate way?>>I think he is almost
saying as well that did we tell the NSA. This metadata might be
more important than data by allowing people to export crypto
because PGP uses rare. Anyone using PGP therefore needs to be
looked at. And when we deregulate a little bit but not
too much, it’s not everywhere, it’s not omnipresent so you can
do a sort and select on people just using crypto for targeting.
That’s a good question and no one here has the answer>>
That’s not why crypto moved out of ITAR. Because we won our case
>>Or he has the answer>>We got the stronger crypto controls
that resulted in export grade encryption back in the 90s. We
got those controls deemed unconstitutional. That’s why it
was — that’s why those controls are slightly less. So the
question is, was the value of meat data part of the reason
that national security establishment in the United
States was okay with that? I think that they weren’t quite
thinking along those lines at that time.>>Next question,
colin?>>AUDIENCE MEMBER: (Off mic)>>Someone has to repeat
that>>Let me just paraphrase into a couple — how did the
crypto regulations affect you, me, in my daily work? And the
short answer is, the crypto regulations probably don’t hurt
my daily work that much because I have already spent enormous
investment in figuring out where those boundaries are and I’m
really comfortable with where — knowing at least where some of
the bright lines are and how I do my work without crossing
them. When it comes to intrusion software, those lines are
inherently a lot more blurry and I think what it will sheen I
spend a lot more time talking to our lawyers at my very generous
University and less time doing my day job, which is filing
grant applications.>>So for Randy, I think the question — I
don’t know if you heard all of that. But I think there is — we
have a lingerie question as to what kind of exceptions are
there or would there be for research use perhaps, on
unintrusion software and those technologies required to build
intrusion software?>>RANDY WHEELER: We are starting at the
point with everything being controlled under the proposed
rule. The possibilities going forward are from my point of
view, endless. They could be certainly a broad license
exception this could be changes to the control language. So it
really depends how the discussion proceeds over the
next few months.>>Thank you>>I’ll just add a quick — even
academics occasionally end up finding themselves on the wrong
end of export control investigation. And it doesn’t
happen that often. But it does happen in very significant ways.
In physics and in bio and to a lesser extent information
systems. I don’t think you can paint it with that specific a
brush.>>Colin is the one person on earth who likes this
thing. So if you want to know more about that position, I
recommend you listen to his Twitter. I think we have time
for one more question. There is two. Two more questions.>>Make
it quick.>>AUDIENCE MEMBER: (Off mic)>>Colin is speaking
on a different Wassenaar panel tomorrow, I think. Tomorrow?
Tomorrow morning.>>You might want to ask why they didn’t
invite any of you guys to comment before they put this
regulation down your throats.>>AUDIENCE MEMBER: (Off mic)>>I
don’t know if you noticed the presence of Randy Wheeler on the
panel. So the question was, are any of us in favor of regulation
at all? And if not, why don’t we have a balanced panel and of
course we have Randy who is the Director of Export.>>But there
is a long discussion about this stuff. Feel free to post to
daily Dave column if you wish to propose things.>>I think that
the point is a valid one that as the software industry continues
to mature, and as a world where we transition more towards a
future cyber war, these technologies are going to or
will become more and more relevant on the battlefield and
there will be increasing government interest not just in
the U.S. but increasing government interest globally in
setting up some kind of regulatory regime. So it
shouldn’t come as a surprised we are here today and I think this
is probably the first of quite a few discussions like this we
will have>>I’ll say I’m not — I’m not sure that I would make a
broad statement saying that no, none of this should ever be
regulated in any way. I can imagine all sorts of bad thing
that could be done with the kinds of software being
discussed here that may well deserve regulation. I don’t know
how to draft regulations without enormous collateral damage.>>
And I would be in favor of regulation that controls the
provision of support for these kinds of technologies to
government end users. That would be a regulation I would get
behind. So, I don’t care about — I don’t care about a remote
administration tool. What I care about is the provision of
support to the domestic version of NSA all across the world.
That should require a license. The tool itself, the technology
behind it, you just go and get that>>Maybe not an export
license. Maybe it should just be something you can sue people in
U.S. court about like we’re doing already and it shouldn’t
be done in export control at>>so with respect to who is for
regulation. It’s worth pointing out as Randy noted earlier,
there will be another round of proposed rules and another
comment period. And I know that the BIS is very interested in
hearing comments from everybody who may be interested in
submitting them and she referenced the number of them
earlier today. So, I’d like to hear her advice on what kind of
comments are most helpful to BIS in figuring out how to do this.
But with the comment about who is for regulation, BIS is not in
the business of making value judgments about whether or not
certain things should be regulated or not it’s there to
fulfill the mission and do the best job it can so comments in
general that are directed to, this is really horrible, go
away, you’re idiots, this is dumb. That kind of thing is not
really helpful obviously. So if you could provide something more
helpful than that to guiding us how to move forward with the
comments.>>RANDY WHEELER: Well, again, the public comment
period for the proposed rule has closed. We certainly will accept
additional public comments but they won’t necessarily be in the
record. But we do want to identify specific issues from
the comments that we received the most important ones, and to
try to flush those out and have all interested parties issues
the ecosystem, the constituencies, including
government agencies, that are involved in cybersecurity, to
weigh in and help us, the inner agency, go forward as
appropriate. Beyond that, I’m not sure that there will be
another proposed rule. It will not be a final rule based on
this proposed rule so there will be an opportunity for more
public comments. We do have the technical Advisory Committee
meetings which we will advertise published in the Federal
Register and we can have open sessions where interested
parties can discuss the issues that have been identified and we
do hope to have broad participation in that process
during the rest of the calendar year. Thank you.>>I have one
more question.>>Hi. There seems to be pretty good
consensus among the panelists on the definitions being not the
best, and ubiquity of some tools and so forth. I wanted to
follow-up on the issue of service being a service provider
and support and the sort of customer that you’re selling
these tools to or people selling the tools to. Nate, one of his
first slide, is it a government enduser or somebody else? Is
there a different regulatory approach that would conceivably
work to focus on who are the buyers and what are they doing
with it? Or do you lose it because if you sell to like an
Ethiopian small businessman that eventually — would be in the
hands of the Ethiopian government?>>That’s a good
point.>>So I think it’s a good question for Randy. If I
understand the question correctly, under a licensing
regime, how do you discern who the end customer is as part of a
licensing process? So if someone is selling to a oppressive
regime or just selling to random interested perhaps researchers
in that same country? Is there any way to distinguish that as
part of the process?>>So the answer is, yes. And we have a
white paper on how to do it. Companies should Intuit a know
your customer policy. We saw this illustrated very nicely in
the hacking team document from metro guard sold to a hacking
team and we saw an e-mail from him to hacking team saying, I
know who your customers are and I’m okay with it. So, that’s the
sort of thing which I would love to bring a lawsuit about. But
yes, a robust know your customer scheme, I think is the best way
to determine it.>>Flowcharts are magic. Have magic powers.>>
Randy, do you have any further comment on know your customer?
>>RANDY WHEELER: That’s just right that is certainly a
provision already in the regulations, the know your
customer, in a licensing process. The end user at times a
statement is required and certainly in a license exception
situation or no license required situation, the know your
customer requirement still applies to ensure a license is
not required. Thanks>>Just one more thing there. I think
honestly, the EFF is going down the wrong path. I’m going to get
him drunk and we’ll correct it. And I’ll tell you why. And it’s
pretty simple, which is that the Wong Wong technology corporation
calls you up and says they want a copy of a random thing, some
gadget ear widget. Now under the current rule set, you’re
supposed to find out if they are owned or mostly owned or
controlled by the Chinese government. But in reality, no
U.S. company can ever really know. There is it no way to
know. So even if you have perfect, and I think immunity
has perfect know your customer abilities, and you have a
flowchart on your wall which explains it to your admin, keep
in mind it’s not a lawyer figuring this out. It’s your
admin. The same person who answers the phone. And they go
through the flowchart and they go, you have a web page. Your
web page looks good. It’s all in Chinese but I don’t know.
Whatever. So I would say dividing a regulatory framework
against this when anyone in China is very difficult to
determine if they are a government-owned,
government-controlled corporation or not is probably
not the right direction to go.>>But the tools we are
concerned about are tools that are sold only tow Governments.
Hacking team and gamma only sell to Governments. So they
certainly know who their customers are>>With that,
unfortunately, we are out of time. The next panel is dying to
get in and play with the AV equipment. So, just want to
extend some recognition here to Mara Tam who did some amazing
things behind-the-scenes to make this panel happen. And also
thank you certainly to Randy Wheeler for really this unique
opportunity I to discuss these proposals with you. And thank
you to all of you for coming in to this talk today. (Applause)

1 thought on “DEF CON 23 – Panel – Licensed to Pwn: Weaponization and Regulation of Security Research

Leave a Reply

Your email address will not be published. Required fields are marked *