September 14, 2017

Are the Shadow Brokers identical with the Second Source?

(Updated: September 23, 2017)

What a lot of people don't know, is that a range of classified documents from the NSA have not been attributed to Edward Snowden, which means that there was at least one other leaker inside the NSA.

Initially, this leaker was called the "second source", and although he was responsible for significant leaks, they got little attention in the US. More media coverage gained the release, since 2016, of NSA hacking tools by the mysterious "Shadow Brokers".

Now, a close look at documents published by the German magazine Der Spiegel in December 2013 provided new indications that the second source could be identical with the Shadow Brokers.

NSA's Cryptologic Center in San Antonio, Texas (2013)
(photo: William Luther - click to enlarge)

The second source

The first leak that was not attributed to Snowden, was of an internal NSA tasking record, showing that German chancellor Angela Merkel was apparently on the NSA's targeting list. The second revelation that was said to come from the same source as the Merkel record, was that of the ANT product catalog, containing a wide range of sophisticated eavesdropping gadgets and techniques.

Security expert Bruce Schneier, who was probably the first to write about the possibility of a second source, said that this source apparently passed his documents to a small group of people in Germany, including hacktivist Jacob Appelbaum and documentary film maker Laura Poitras.

Because Poitras also received one of the initial sets of documents from Snowden, it is sometimes assumed that the documents from the Second Source may actually stem from the Snowden trove, despite not being attributed as such. For some of the individual documents this was contradicted by Glenn Greenwald and Edward Snowden though.

Spiegel reportings

The ANT catalog was published by the German magazine Der Spiegel on December 29, 2013. The original article was in German and written by Jacob Appelbaum, Judith Horchert, Ole Reißmann, Marcel Rosenbach, Jörg Schindler and Christian Stöcker. A translation in English mentioned the names of Jacob Appelbaum, Judith Horchert and Christian Stöcker.

Although this catalog got most of the attention, not at least because Appelbaum explained the various tools during a presentation at the hackers conference CCC on December 30, it was actually just an addition to Der Spiegel's extensive main piece about the hacking division of the NSA, called Tailored Access Operations (TAO).

This article was written by Jacob Appelbaum, Marcel Rosenbach, Jörg Schindler, Holger Stark and Christian Stöcker, with the cooperation of Andy Müller-Maguhn, Judith Horchert, Laura Poitras and Ole Reißmann. There was also a translation in English prepared by the Spiegel staff based upon reporting "by Jacob Appelbaum, Laura Poitras, Marcel Rosenbach, Christian Stöcker, Jörg Schindler and Holger Stark."

TAO documents

This main piece was accompanied by various NSA documents: one slide about FOXACID, a partial presentation about QUANTUM, two separate pages from other documents, as well as complete powerpoint presentations about QUANTUM tasking, the TAO unit at NSA/CSS Texas, and the QFIRE architecture:

(click to go to the various documents)

Not Snowden?

Apparently never noticed before, is that not only the ANT product catalog, but also these other presentations and documents were not attributed to Snowden. In both the German and the English version, the whole lengthy article contains multiple times phrases like "internal NSA documents viewed by SPIEGEL" but never in combination with the name of Edward Snowden.

This is remarkable, because for the media, it's usually almost some kind of honor to publish documents provided by Snowden, which is then clearly mentioned in their reporting. In those cases, the byline includes the name of the one who actually provided the documents on Snowden's behalf, often Glenn Greenwald and for Der Spiegel, Laura Poitras.

But both articles from December 29 have Jacob Appelbaum, instead of Poitras in the byline, which seems to be an indication that here, the top secret NSA documents were provided by Appelbaum, likely as the middleman for the mysterious second source.

Exception: FOXACID slide

There's one exception though: the description of the FOXACID slide says that it is from an NSA presentation from the Snowden cache - this was confirmed when on August 19, 2016, The Intercept eventually published the full presentation about FOXACID.

This slide was probably provided by Laura Poitras, from her cache of Snowden documents, which would explain why she was mentioned as one of the persons that provided assistance for Der Spiegel's main piece of December 29.

The other presentations have not been published as part of the Snowden revelations, there's only one with a similar layout (from Booz Allen's SDS unit), but is about a different topic.


If not only the ANT Product Catalog, but also these other NSA presentations about the TAO division were not provided by Snowden, but by the second source, what's the significance of that?

Analysing the range of revelations that were not attributed to Snowden, resulted in the following list of documents that were likely leaked by the second source:

- Chancellor Merkel tasking record
- TAO product catalog
- XKEYSCORE rules: New Zealand
- NSA tasking & reporting France, Germany, Brazil, Japan
- XKEYSCORE agreement between NSA, BND and BfV(?)
- NSA tasking & reporting EU, Italy, UN

Except for the TAO catalog, one of the things that all these documents have in common, is that they are different from the usual powerpoint presentations, program manuals and internal wiki pages that make up the biggest part of the Snowden revelations.

(Of course, absence of evidence is no evidence of absence, but as these second source documents are often more significant than many other Snowden files, there seems to be no reason not to publish them)

The additional December 29 files do actually fit the typical sort of documents from Snowden, which makes it more difficult to distinguish between documents from Snowden and those from the other leaker(s).

The Shadow Brokers

If we look at the content of the files, we see that those from Der Spiegel's December 29 article are all about NSA's hacking operations. There have been several Snowden stories about that topic, but more spectacular became the release, since August 2016, of actual NSA hacking tools by a mysterious person or group called The Shadow Brokers (TSB or SB).

There has been a lot of speculation about who could be behind this and how he, she or they got access to these sensitive files. One option is an NSA insider, either on his own, in cooperation with crypto-anarchists, or as a mole directed by a hostile intelligence agency.

Another suggestion was that an NSA hacker mistakenly uploaded his whole toolkit to a server outside the NSA's secure networks (also called a "staging server" or "redirector" to mask its true location) and that someone was able to grab the files from there - this option was for example favored by Snowden.


The latter theory was falsified when on April 14, 2017, the Shadow Brokers did not only publish an archive containing a series of Windows exploits, but also several documents and top secret presentation slides about NSA's infiltration of the banking network SWIFT - things unlikely to be on a staging server, which makes that the source behind the Shadow Brokers is most likely an insider.

On July 28, the website CyberScoop reported that as part of their investigation into the Shadow Brokers leaks, US government counterintelligence investigators contacted former NSA employees in an effort to identify a possible disgruntled insider.

(just a few days ago, the Shadow Brokers released a manual for the hacking framework UNITEDRAKE, strangely enough without date and classification markings, but again something that one wouldn't find on an outside staging server)

Same source?

With the documents published by the Shadow Brokers apparently being stolen by an insider at NSA, the obvious question is: could the Shadow Brokers be identical with the Second Source? (see update)

One interesting fact is that the last revelation that could be attributed to the second source occured on February 23, 2016, and that in August of that year the Shadow Brokers started with their release of hacking files. This could mean that the second source decided to publish his documents in the more distinct and noticeable way under the guise of the Shadow Brokers.

But there's probably also a much more direct connection: the batch of documents published along with Der Spiegel's main piece from December 29, 2013 include a presentation about the TAO unit at NSA's Cryptologic Center in San Antonio, Texas, known as NSA/CSS Texas (NSAT):

TAO Texas presentation, published by Der Spiegel in December 2013
(click for the full presentation)

And surprisingly, the series of three slides that were released by the Shadow Brokers on April 14 were also from NSA/CSS Texas. They show three seals: in the upper left corner those of NSA and CSS and in the upper right corner that of the Texas Cryptologic Center:

TAO Texas slide, published by the Shadow Brokers in April 2017
(click for the full presentation)


It's quite remarkable that among the hundreds of NSA documents that have been published so far, there are only these two sets from NSA/CSS Texas, which is responsible for operations in Latin America, the Caribbean, and along the Atlantic littoral of Africa in support of the US Southern and Central Commands.

Update: The three Shadow Brokers slides from NSA/CSS Texas show operations against EastNets and Business Computer Group (BCG), which are both Service Bureaus for the banking network SWIFT. EastNets has offices in Belgium, Jordan, Egypt and UAE and was targeted under the codename JEEPFLEA_MARKET, while BCG serves Panama and Venezuela and was targeted under JEEPFLEA_POWDER.

Besides the one in San Antonio, Texas, NSA has three other regional Cryptologic Centers in the US: in Augusta, Georgia, in Honolulu, Hawaii and in Denver, Colorado. These four locations were established in 1995 as Regional Security Operations Centers (RSOC) in order to disperse operational facilities from the Washington DC area, providing redundancy in the event of an emergency.

So far, no documents from any of these regional centers have been published, except for the two from NSA/CSS Texas. This could be a strong indication that they came from the same source - and it seems plausible to assume that that source is someone who actually worked at that NSA location in San Antonio.


This person may only have stolen files that were available at his own workplace, as it should be realized that not every leaker necessarily has similar broad access like Snowden had (and gained) in his job as a systems administrator.

Snowden on the other hand may only have downloaded things from an intranet for NSA as a whole (assuming that would contain the most interesting files) and leaving the local network for his Hawaii office untouched - which would explain why we never saw any documents marked NSA/CSS Hawaii (another reason could be that such documents would have made it easier to identify him).

Given the many hacking files, it's tempting to assume that the second source/Shadow Brokers was an NSA hacker at the Texas TAO unit. It's not clear though whether someone in such a position would also have had the access to the intelligence reports and traditional tasking lists which were published by Wikileaks. It's also possible that those documents came from a different source.


One final thing that the revelations from the second source and the Shadow Brokers seem to have in common is the motivation: none of their documents reveal serious abuses or illegal methods, but only compromise methods and operations, and discredit US intelligence.

Most of these documents weren't vetted by professional journalists either: although initially published by Der Spiegel and some other German media, later files were made public by the uncritical website Wikileaks, while the Shadow Brokers postings come without any intermediary on sites like Pastebin, Medium and Steemit.

(In March 2017, Wikileaks started the "Vault 7" series in which they publish secret hacking tools from the CIA. These files have dates between November 2003 and March 2016 and are therefore more recent that those from the Shadow Brokers, with their newest files being dated October 18, 2013 - some 5 months after Snowden left the NSA and around the same time when Der Spiegel published the first document from the second source)


On the weblog there are some additional thoughts about this issue: the author is, for various reasons, skeptical about the Shadow Brokers being a disgruntled NSA employee or contractor, and therefore that he could be identical with the Second Source. As an alternative, EmptyWheel suggests that Jakob Appelbaum and the Shadow Brokers may have a mutually shared source.

I can agree with that, as I may not have made clear enough that the Second Source is the person who was able to actually steal documents from inside NSA, while the Shadow Brokers is a group or a single person who is responsible for publishing the files, just like Der Spiegel and Wikileaks did for most of the documents attributed to the Second Source.

Of course it would be possible that the Second Source eventually started to publish his documents himself under the covername Shadow Brokers, but as noted by EmptyWheel, there are several indications that makes this less likely.

A slightly different option is that the Second Source provided his documents to Jacob Appelbaum and that he had them published by Der Spiegel and Wikileaks, and that later on, either Appelbaum himself acted as the Shadow Brokers, or gave the files to someone else operating under that guise.

Links and sources
- UNITEDRAKE and hacking under FISA Orders
- Shadow Brokers' Persistence: Where TSB has signed, message, hosted, and collected
- The US Intelligence Community has a Third Leaker

August 13, 2017

Do NSA compliance reports point to an unknown classification compartment?

(UPDATED August 14, 2017)

On July 12, the American Civil Liberties Union (ACLU) published two Top Secret NSA compliance reports, which were obtained after declassification under the Freedom Of Information Act (FOIA). Here we will take a look the classification marking of these reports, of which one part has been redacted:

Both documents are quarterly reports from the NSA to the Foreign Intelligence Surveillance Court (FISC) on compliance under Section 702 of the FISA Amendments Act (FAA), which governs both the PRISM and the Upstream collection efforts. One of the reports is from March 2014, the other one from March 2015.

More about the content of these two compliance reports can be found in this article by The Hill, as well as in this posting on the weblog Here we will take a look at the classification of these reports.

Classification marking

The classification line of both reports is: TOP SECRET//[...]/SI//ORCON/NOFORN, which stands for:

- TOP SECRET: the highest classification level
- [...]: an unknown SCI control system
- SI: the SCI control system "Special Intelligence"
- ORCON: the dissemination restriction "Originator Controlled"
- NOFORN: the dissemination restriction "No Foreign Nationals"

As we can see, the double slash behind TOP SECRET is followed by a short space that has been redacted. Then comes a single slash, followed by SI for the control system for Signals Intelligence (SIGINT), which was formerly denoted as COMINT. This means that the redaction also hides a so-called control system, used for protecting national intelligence information concerning sources and methods.

The semiannual Section 702 FAA compliance reports, like this one from 2015, which are prepared by the Attorney General and the Director of National Intelligence, don't have the redacted marking and are just TOP SECRET//SI//NOFORN.


Meanwhile, a reader suggested that the redacted part of the classification line might hide HCS, which is the abbreviation for HUMINT Control System. HUMINT stands for Human Intelligence, and therefore, HCS is the control system that protects intelligence from the CIA. And indeed, it appears that "HCS" fits the redacted space perfectly:

HCS itself isn't classified, so the reason why this marking appears redacted here, is probably that NSA only declassified its own part and redacted everything related to the CIA. Analysts of the CIA request and receive information from 702 FAA collection, but the scope of their involvement remains classified.

With HCS being the most likely option for the redacted space in the 2014 and 2015 compliance reports, please read the following with that in mind!


Another option that comes to mind is the STELLARWIND compartment, which was created in October 2001 to protect NSA's new collection methods as authorized by president George W. Bush. This is officially known as the President's Surveillance Program (PSP) and more popular as the "warrantless wiretapping".

For classification purposes, the abbreviation of STELLARWIND is STLW, which is too long for the redacted space in the marking on the compliance reports. The STELLARWIND classification guide from January 2009 does provide an interesting alternative though, where it says:

"The markings "TSP" and "Compartmented" were at times used in briefing materials and documentation associated with the STELLAR WIND program. "TSP" and "Compartmented" were used primarily by the National Security Agency (NSA) Legislative Affairs Office (LAO), NSA Office of General Counsel (OGC), and the Executive Branch in briefings and declarations intended for external audiences, such as Congress and the courts. The term "TSP" was initially used in relation to only that portion of the Program that was publicly disclosed by the President in December 2005. These markings should be considered the same as the STELLARWIND marking, but should not be directly associated with the program."

In several documents that had been presented to the FISA Court and meanwhile have been declassified by the US government, we can see this TSP marking:

Classified declaration of NSA director Alexander, April 20, 2007.

The two recently declassified compliance reports from 2014 and 2015 were also meant for the FISA Court, and if we try out "TSP", it fits the redacted space remarkably well:


PSP/TSP collection (2001-2007)

Under the President's Surveillance Program (PSP), as protected by the STELLARWIND compartment, it became possible for NSA to not only collect fully foreign communications, but also those with just one end foreign - the express aim was to find foreign terrorists with connections inside the US. Under the PSP the following data were collected (with their succeeding legal authorizations):

- Telephony content (since 2008: Section 702 FAA Upstream collection)
- Internet content (since 2008: Section 702 FAA Upstream collection)

- Telephony metadata (2006-2015: Section 215 (BR/FISA) bulk collection)
- Internet metadata (2004-2011: Section 402 (PR/TT) bulk collection)

The bulk collection of internet metadata was brought from the president's authority under that of the FISA Court (FISC) in July 2004, and the same happened with the bulk telephone metadata in May 2006.

The collection of both telephone and internet content became also authorized by FISC orders as of January 2007, which was temporarily replaced by the Protect America Act (PAA) in August 2007 and then permanently by Section 702 of the FISA Amendments Act (FAA) in July 2008.

Section 702 FAA is also the legal foundation for PRISM collection, which started in September 2007 with data being provided by Microsoft. Until October 2012, another eight internet companies had followed. While Upstream collection, at major telecom switches, only provides future communications in transit, PRISM gives access to stored data from a target's past too.

After revelations by the New York Times in December 2005, president Bush admitted that NSA was collecting the one-end foreign telephone and internet content and named it the Terrorist Surveillance Program (TSP). Bush stayed silent though about the other part of the PSP, which involved the bulk collection of domestic metadata. This came to light in June 2013, when Snowden provided the Verizon order to the press.

Another option

As we have seen, the TSP marking fits the redacted space in the classification line of the compliance reports very well, but of course it's always possible that a different abbreviation might be hidden there. In documents that have been declassified earlier, "TSP" was not redacted, so strictly spoken, it shouldn't have been redacted here.

And there's indeed another option: on September 6, 2014, the US Justice Department released a declassified version of a 2004 memorandum about the STELLARWIND program. The classification line of this document has a similar short redaction right after "Top Secret", just like in the compliance reports:

Classification marking of the 2004 DoJ memorandum about STELLAR WIND

Interestingly, "TSP" would also fit the redacted space here - but this wouldn't make much sense, as TSP was meant as a replacement for STELLARWIND for audiences who didn't have a need-to-know for the STELLARWIND cover term - and "STELLAR WIND" is in the classification line here too. Also, the name Terrorist Surveillance Program probably didn't exist in 2004, as it was apparently first used by President Bush in a speech on January 23, 2006.

So, it's not very likely that "TSP" is the marking that was redacted in the 2004 memorandum, but position and length of the marking indicate that it's very well possible that it's actually the same control system as in the classification line of the compliance reports from 2014 and 2015 - with an abbreviation of almost exactly the same length as "TSP".

Regarding the status of this mysterious marking: in the 2004 document it's shown between double slashes, which is strange, because according to the official classification manuals, there cannot be something between two double slashes in that position (see the chart below).

If this double slash is correct, then we would have a complete new category which isn't in the (public) classification manuals. This reminds of the UMBRA marking, which also appeared unexpectedly between double slashes in a classification line.

Another option is that the double slash behind the redacted marking is actually a mistake and there should have been just a single slash, just like in the classification lines of the 2014 and 2015 compliance reports. In this case, the marking represents a normal control system like SI, HCS, and several others mentioned in the classification manuals.

Overview of the categories and formatting for the US classification and control markings
From the Intelligence Community Classification Manual 6.0 from December 2013
(click to enlarge)

July 14, 2017

Dutch report provides metadata numbers to compare with Snowden documents

(Updated: September 21, 2017)

Since the Snowden revelations, we know that signals intelligence agencies are trying to acquire large sets of telephone metadata in order to analyse them in support of protecting their national security.

Less known is that commercial companies also analyse similar big data sets, albeit for research purposes and with personal information being anonymized.

Now, a research report from the Netherlands provides us with actual numbers of mobile telephone metadata, which can be used to compare with the numbers that NSA and GCHQ collected according to the Snowden documents.

Tourist movements

Recently published was a report about visitor movements in and around the Dutch capital of Amsterdam. It was prepared by the economic research company Decisio on behalf of the province of Noord-Holland and the municipalities of Amsterdam and Zandvoort.

Since a few years, Amsterdam almost suffers from a huge increase of tourists, but it was difficult to get detailed insights in where they come from, where they stay and which areas of the city, as well as which surrounding towns are most popular.

Now, these insights became available by using information from the "tracking device" carried by almost every individual: the mobile phone.

Anonymized data sets

Decisio acquired a huge set of mobile telephone metadata from Vodafone, which is the second largest provider in the Netherlands, with over 5 million customers. When they use their mobile phones, they connect to one of the 32.000 cell towers or base stations, which associate the phone number with a location.

Each month, Vodafone provides these data to another research company called Mezuro, which processes and analyses them to map the movements using a grid of 1250 regions containing multiple base station cells. The results were then analysed by Decisio and compared with other information sources.

But before that, the Vodafone metadata were anonymized by replacing every phone number with a random number that changes every month. Foreign phone numbers were replaced daily. Also, only the movements of groups of more than 15 numbers were reported, so it's impossible to track the movements of individual phone users.

Development of the average number of daily mobile phone transactions
from Vodafone users between January 2013 and September 2015
(source: Decisio research report)

Mobile phone metadata

Most interesting parts of the report are the details about the telephone metadata: Mezuro periodically receives information about some 3 million Vodafone phone numbers that are active on a daily basis. These phones generated 400 million "transactions", or Call Detail Records (CDRs) a day.

These transactions are the moments that a mobile phone connects to a cell tower, not only for a phone call or a text message (SMS), but increasingly often for a social media posting, sending or receiving an e-mail, a Google search or checking a website - for Dutch users, this is on average 100 times a day.

An article from October 2013 about Mezuro says that the company analyzed some 150 million data points daily and that an average smartphone connects 150 to 200 times a day with a cell tower.

This number was confirmed during a parliamentary hearing in Germany, when someone from BND explained that one cell phone generates between 100 and 200 metadata and business records a day.

Update: similar numbers come from the Russian billing (and interception) company Peter-Service, which installs billing installations capable of 18 million subscribers creating 2,851 billion transactions a day, which equals 158 transactions per person per day.

If we take these metadata as the rows of a (database) table, each of them contain multiple fields, corresponding to columns for information pieces like for example the number calling, the number called, date, time, cell tower location, and information needed to transfer various types of messages.


For the tourism research, the Vodafone data were multiplied in order to get the numbers for the full population. The multiplier changes daily depending on the day of the week, holidays, etc, but lies roughly around 5 (for foreign visitors it's much more difficult to calculate this number).

As this total also includes people who don't use a mobile phone, the multiplier for the total number of metadata must be lower. According to the report, the users of the Vodafone network account for 1/3 of all mobile phone users in the Netherlands, so here we can use 3 as multiplier.

That makes that in 2016, all Dutch mobile phones generated some 1200 million transactions a day. In a month that's over 36 billion and in a year 432 billion telephone metadata records.

For comparison with the numbers from the Snowden documents, we have to look for the numbers from early 2013. The chart from the report shows that in January 2013, there were ca. 85 million transactions by Vodafone users a day, which makes 255 million for all Dutch users. In a month that's 7,65 billion.


Now, let's take a look at some of the numbers from the Snowden revelations. For the Netherlands there was a chart from the NSA tool BOUNDLESSINFORMANT, which shows 1,8 million telephone metadata records for 30 days around January 1, 2013.

Initially it was thought that this were Dutch data sucked up by NSA, but later it came out that they were actually collected by Dutch military intelligence, most likely in Afghanistan, and subsequently shared with the Americans.

Now that we know that in the same period of time, the Dutch mobile phone users alone already accounted for over 7 billion metadata, 1,8 million is a tiny number, maybe generated by not more than 2500 smartphones. In Afghanistan, old fashioned cell phones may have created less transactions, so the 1,8 million metadata could have been the traffic captured from a small town.

Update: On Twitter, a Dutch journalist involved with the Snowden revelations said that the 1,8 million records represent some 12 million pieces of metadata (which means one record consists of at least 6 fields) and that the Dutch Ministry of Defence had confirmed that they were collected from Somalia.

The BOUNDLESSINFORMANT chart for the Netherlands with data
collected from December 10, 2012 to January 8, 2013
(click to enlarge)

Late 2013, major European newspapers published similar charts for other countries too, again claiming that they showed how many phone calls NSA was intercepting. But even if those claims were true, the 70 million the BOUNDLESSINFORMANT chart presented for France, 60 million for Spain, 45 million for Italy and 33 million for Norway, are tiny numbers given the actual 7,65 billion metadata for a small country like the Netherlands.

Even the 552 million metadata in the chart for Germany doesn't come close. If the Netherlands with some 16 million people generated 7,65 billion mobile phone metadata a month, then for 80 million German citizens that number would be over 38 billion.

And to be clear: the data represented in these specific BOUNDLESSINFORMANT charts were not collected by NSA in Europe, but shared with NSA by European intelligence agencies, as part of their military cooperation in various crisis zones.

NSA and GCHQ totals

Finally, we can look at how many telephony metadata NSA and GCHQ collect in total and compare that with the numbers from the Netherlands. In 2012, the British GCHQ "was handling 600m "telephone events" each day" - according to Snowden documents seen by The Guardian.

This seems a surprisingly small number compared to the 225 million transactions generated by Dutch users, but it's possible that the 600 million only apply to traditional telephone and SMS metadata, excluding the internet data from smartphones.

The NSA collected a total of 135 billion telephone metadata a month during the first half of 2012. This is some 17 times the amount for the Netherlands as a whole - again not a very excessive number, as it would roughly equal the telephone metadata of around 300 million people, which is more or less the population in the Middle East.

The volumes of NSA metadata collection between January and June 2012
(click to enlarge)


During the Snowden revelations, the press was eager to present numbers about NSA and GCHQ data collection that seemed impressingly high. But not a single media outlet took the time or effort to come up with the total numbers of telephone and internet communications, needed to put them in the right perspective.

From the research report about Amsterdam tourism we finally learned what the actual number of mobile telephone metadata for a western country look like. Although we still don't know how exactly NSA and GCHQ are counting their metadata, comparing them to the numbers from the Netherlands shows that their collection efforts may be not as excessive as initially thought.

Links and sources
- Decisio: Bezoekersstromen Amsterdam - Zandvoort (2017)
- Autoriteit Consument & Markt: Telecommonitor eerste kwartaal 2016 (2016)
- ITU: Innovation of tourism statistics through the use of new big data sources (2014)
- CBS: Rapportage project impact ICT; Mobiele telefonie (2013)

One interesting result from the tourism report is that measuring the number of visitors of Amsterdam's annual Gay Pride showed that instead of the 560.000 visitors according to the organisation, only 115.000 visitors came from outside the city center, additional to the 235.000 people who are present on every Saturday and may or may not have watched the event. This confirms that visitor numbers for free public events are often significantly exaggerated.

June 8, 2017

Dutch-Russian cyber crime case reveals how the police taps the internet

(Updated: August 26, 2017)

About how signals intelligence agencies, like NSA and GCHQ, are intercepting communications, we learned a lot from the Snowden revelations and the German parliamentary inquiry, but also from new legislation in France, the Netherlands and the United Kingdom.

Much less is known about the practice of tapping by law enforcement, like for example the FBI and police forces. Now, a case from the Netherlands provides some interesting insights in how Dutch police intercepts internet communications - in a way that comes remarkably close to the bulk collection by intelligence agencies.

Office of the Team High Tech Crime (THTC) of the Dutch police in Driebergen
(photo: NRC/Merlin Daleman)

Cooperation with the Russians

On Saturday, May 27, the Dutch newspaper De Volkskrant came with a surprising story about the cooperation between the Team High Tech Crime (THTC) of the Dutch police and officials from the Russian federal security service FSB, which is the main successor to the notorious KGB.

Since 2009, regular meetings are held in the Netherlands, in which also officials from the FBI participate. The aim is to cooperate in tracking down and eventually arresting cyber criminals. The Volkskrant's front page report is accompanied by an extensive background story, which contains some more worrying details, but is only available in Dutch.

The cooperation with the Russians dates back to September 2007, when the head of THTC attended a conference in the Russian city of Khabarovsk, at which CIA, FBI, Mossad, BND and other agencies were present. The head of THTC was able to create a connection to the FSB and their deputy head of the Center for Information Security (TsIB), Sergei Mikhailov, became the liaison for the Dutch police and would regularly visit the Netherlands.

Meetings in Driebergen

Initially, the meetings with the Russians were held in the Dutch village of Driebergen, where the Team High Tech Crime has its offices. The Dutch security service AIVD was apparently not very fond of this, so every visit of for example Mikhailov had to be reported, and since 2012, every police officer who had contact with someone from the FSB was briefed by the AIVD before and after every meeting.

The FSB, much like the FBI, isn't just responsible for law enforcement, but is also Russia's secret service for domestic security. This made AIVD worried that FSB officers could use their visits to the Netherlands for spying - although strictly spoken, collecting foreign intelligence is the task of another Russian agency, the SVR.

The police compound in Driebergen started as highway patrol station, but nowadays houses some of the most sensitive units of the Dutch police, including the national criminal investigation branch and the Unit Landelijke Interceptie (or Lawful Interception, ULI; nowadays: Interceptie & Sensing, I&S), which was created in 2005 as the central facility for internet tapping, as well as for telephone tapping on behalf of all the smaller police districts.*

The police compound in the village of Driebergen
(photo via Flickr)

Security incident

There was at least one security incident in Driebergen: De Volkskrant describes that during a meeting with FBI and FSB, a Russian official came to a member of the Dutch police team, pointed at someone from the FBI and said "he is copying your data". An investigator went looking and saw that indeed the American had a thumb drive in a police laptop and was copying Dutch information. Whether this had any consequences was not reported.

In 2014, the cooperation with Russia came under pressure: in July, there was the Russian annexation of the Crimea and shortly aftwerwards, flight MH17 was shot down, killing 193 Dutch citizens. The criminal investigation of this case also takes place in Driebergen, so the police decided to move to meetings with FSB officials from Driebergen to police stations in Amsterdam and Rotterdam.


Intercepting at Leaseweb

The first case in which Dutch police and Russian FSB cooperated started in 2008, when Russian criminals used the ZeuS trojan horse malware to spoof the login screen of banks in order to capture user credentials, and steal the money from bank accounts without a trace.

Often these criminals used servers of the Dutch hosting company Leaseweb, which offers relatively anonymous and cheap services as well as high-speed connections, as it is close to the large Amsterdam internet exchange AMS-IX. To communicate with eachother, the criminals used the messenger service ICQ, which is still popular in Russia and Eastern Europe, but doesn't use encryption.

To catch the criminals behind the ZeuS malware, the Dutch police team set up operation Roerdomp (the Dutch name for the Eurasian bittern) and in October 2008, they asked other countries for the ICQ numbers of known cyber criminals. Within 3 months, authorities from the US, Germany, Britain, the Ukraine and Russia provided a total of 436 ICQ numbers. In January 2009, the public prosecutor and an examining judge approved the interception of communications associated with these numbers.

ICQ logo and interface

DPI filtering

To acquire these ICQ communications, the police had decided to intercept all ICQ traffic from Russia that went through the Leaseweb servers. For that purpose they bought equipment for deep-packet inspection (DPI) worth 600.000,- euro.

DPI devices are able to examine the packets that make up internet traffic and filter them according to predefined criteria, usually to prevent viruses and spam, but in this case for intercepting communications.

High-end DPI equipment, from manufacturers like Narus and Verint, can also recreate ("sessionize") the communication sessions in order to filter complete files and messages out - which is also one of the main features of NSA's XKEYSCORE system.

The Volkskrant reports that after the interception was approved, the new equipment was connected to the servers of Leaseweb, but actually, Leaseweb will have splitted the traffic on its main backbone cable, creating a copy of all the data, which was then directed to the police computer - telecom and internet companies really don't like outsiders to install equipment onto their actual networks.

Next, all the copied Leaseweb traffic, some 50 Gigabit per second for 4 to 10 million websites, went through the DPI machine. First the police filtered out all ICQ traffic, and then the ICQ traffic associated with the list of the 436 selected numbers. This went on for 3 months, so the warrant was apparently renewed a few times, as an approval for targeted interception is initially limited to a period of 4 weeks.

Update: On July 5, 2017, it was reported that in the brand new Equinix AM4 data center in Amsterdam (with over 120.000 servers, connected to 150 networks), there's a highly secure section which is used by the Dutch government - could that be intended for intercepting the servers that (foreign) companies are hosting there?

Leaseweb headquarters in Amsterdam
(click to enlarge)

Some questions

The description of the tapping operation by De Volkskrant raises some questions. Government filtering systems having access to all the internet traffic of a company is the way that (signals) intelligence agencies are conducting bulk collection, not the way that law enforcement is supposed to do targeted interception.

In western countries, the police is generally only allowed to tap communications associated with individually identified suspects or specific communication identifiers, like phone numbers and e-mail addresses. In the ZeuS case, it was probably argued that it was targeted interception because there were 436 specific identifiers: the ICQ numbers of known cyber criminals.

Foreign selectors

First, this case immediately reminds of the selector affair that came to light through the German parliamentary inquiry into the cooperation between NSA and BND. For years, NSA provided the Germans with millions of internet identifiers, which they entered into their satellite collection system, without being able to see to whom these identifiers belonged.

Could that have happened to the Dutch police too? Were they able to verify that each one of the 436 ICQ numbers was used by a cyber criminal, or did they just trusted the foreign authority that provided them?

For this kind of international cooperation, it's often inevitable that you have to trust your foreign partners, but then you should also try to make sure that the data collection is as careful and targeted as possible.

Dutch internet tapping

One way to assure that is through technical means. For telephone tapping this is relatively easy, because telephone switches have built-in tapping capabilities based upon international standards. For internet tapping this is different and external devices have to be used to pick out the communications of interest.

In the Netherlands, the interception of internet data uses the Transport of Intercepted IP Traffic (TIIT) protocol, which ensures that the police only gets the internet data associated with an IP or e-mail address for which there's a warrant (managed through the Warrant Management System, WMS).

Overview of the TIIT protocol for IP and e-mail interception
(click to enlarge)

First, an Internet Service Provider (ISP) copies all its traffic and leads the copy to a secured interception network on its own premises. There, a sniffer machine (S1) filters out the data that have to be intercepted, and encrypts these with a key that is associated with a particular warrant.

Then, these data go to the ISP collector machine (S2), which sets up a connection, through an encrypted tunnel over a regular internet link, to a government collector machine (T1), which receives the data from one or more S2 machines.

The T1 devices are managed by the central interception unit in Driebergen and from there, the intercepted data are distributed to computers (T2) at the tapping rooms (tapkamers, nowadays: BOB-kamers) of the police districts. There, they are stored and decrypted so the intercepted communications become available in plain text.

Intercepting hosting providers

With the TIIT protocol, the police doesn't get access to the copy of an ISP's entire traffic: it's the ISP that controls the sniffer machine that filters out the communications that belong to a particular suspect. But at Leaseweb it was apparently the police that controlled the sniffer (in the form of DPI equipment) where all the traffic passed through.

The most likely reason for this is that Leaseweb is a hosting provider and it's considered that such companies don't have to comply with the Dutch Telecommunications Law that says that public communication networks or services have to be interceptable. Therefore, hosting providers were not required to install the tapping facilities like the telephone and internet access companies have.

But the hosting companies can of course cooperate voluntarily when the police presents them a warrant. However, when the new Secret Services Act comes into force, such non-public communication providers do have to tolerate interception on behalf of AIVD and MIVD, but they don't need to have pre-installed tapping equipment.

This means that in both cases, even for targeted interception, the government will control the sniffer equipment for filtering up to a company's entire traffic - something that digital rights groups like the ACLU already consider to be unlawful "bulk surveillance."


Another question is how to make sure that the police doesn't misuse it's power when for example a hosting provider voluntarily provides access to their entire traffic. Maybe the police has internal protocols for that, but while interception conducted by the secret services is subject to independent oversight, police tapping is not.

It's considered that in criminal cases, a judge will eventually decide whether certain police methods are lawful or not, but in practice, judges often lack the necessary technical knowledge, while police and public prosecutors try to hide these sensitive techniques. It's not clear whether any suspect in the ZeuS case was tried before a Dutch court.

Untargeted interception

The ZeuS case shows that not only the networks of telecommunications and internet service providers can be useful to intercept, but also hosting providers like Leaseweb, especially when their servers are used by foreign companies to host their internet (communication) services - useful, not only for the police, but also for the secret services AIVD and MIVD.

Soon, both services can even go a step further, as the new Secret Services Act will also allow them to conduct untargeted cable interception. That means that they may not only filter out communications that are associated with already known identifiers, but also (temporarily) store all the metadata and a lot of content in order to search for data that belong to yet unknown targets.

In the public debate about the new law, there was a lot of speculation about how the new untargeted cable access will be implemented, but the interception at Leaseweb, as described by De Volkskrant, gives a very concrete example of what can be expected.

National watch center of the Royal Marechaussee in Driebergen
with a large dark gray Philips PNVX crypto telephone

The end of ZeuS

After collecting the messages associated with the 436 ICQ numbers and subsequently analysing them, it came out that one particular ICQ number acted as the leader of the cyber crime network. In one of the intercepted conversations this person even admitted to be the designer of the ZeuS malware.

The police gave him the codename "Umbro", but he himself used aliasses like Lucky12345, Monstr, Slavik, IOO, Pollingsoon, and Nu11. De Volkskrant story doesn't tell how the police found out the real identity of "Umbro" and it was only in 2014, under the international law enforcement Operation Tovar, that he was identified as Evgeniy Mikhailovich Bogachev, born October 28, 1983.

Already in 2013, investigators noticed that the ZeuS virus wasn't just used for stealing money anymore, but also for finding out very specific information about government officials of Russia's neighbours. Dutch police and the FBI became convinced that "Umbro" (Bogachev) had started working for Russian intelligence too.

To be or not to be arrested

The latter seems to be one of the reasons that, after the hack of the Democratic National Committee (DNC) in 2016, the US government put Bogachev on a list of sanctioned individuals. Besides that, his malware was also responsible for stealing over 100 million USD from American organizations. However, Bogachev is still at large, probably because he is useful for Russian intelligence operations.

For the Dutch police team there was another unpleasant surprise: Sergei Mikhailov, the FSB officer who had become such a familiar face for them, was suddenly arrested in December 2016 - according to Russian press reports because he and Kaspersky expert Ruslan Stojanov had leaked information to US intelligence.

Nobody knows whether this is true or where Mikhailov is now, but the cooperation between Dutch police and the Russian FSB continues.

In August 2017, Russian media reported that Sergei Mikhailov and his deputy Dmitry Dokuchaev were charged with treason after they were found to have helped the CIA catch two notorious Russian hackers: Roman Seleznev, who was arrested in 2014 on the Maldives, and Yevgeniy Nikulin, who was arrested in the Czech Republic in 2016.

Links and sources
- Moscow's cyber-defense How the Russian government plans to protect the country from the coming cyberwar (2017)
- Dutch police works together with Russia's FSB, despite political tensions (2017)
- Russen schakelden contactpersoon van Nederlandse cyberpolitie uit (2017)
- Inspectie V en J: Meldkamer Landelijke Eenheid Politie (.pdf) (2014)
- Ars Technica: Deep packet inspection meets ‘Net neutrality, CALEA (2007)
- Aftapbaarheid van telecommunicatie (.pdf) (2005)
- Geschiedenis AVD Driebergen (2002)