Month: July 2015

Ten standards for oversight and transparency of national intelligence services: custodiet ipsos custodes

ten-standards

On July 23rd 2015, legal scholars from the Institute for Information Law of the University of Amsterdam — Sarah Eskens, Ot van Daalen (@DigiDefence) and Nico van Eijk — published a report (.pdf, in English) entitled “Ten standards for oversight and transparency of national intelligence services”. The proposed standards are substantiated by, among others, analysis of CJEU jurisprudence.

The report appears three weeks after the Dutch government released a new intelligence bill into public consultation (details). That bill includes significant expansions of power, notably enabling bulk interception of domestic and foreign cable communications, and mandatory cooperation from (to be selected) “providers of communication services” — a broad category that covers internet providers, hosting providers (cloud? CDNs?), and website operators. Notably, the authors of the report plead for ex ante review of interception and intelligence operations by a court. The bill neither includes independent ex ante review, nor any court involvement.

The executive summary from the report:

Executive summary

The main goal of this report is to contribute to the policy debate on surveillance by intelligence services from the perspective of oversight and transparency. Both are considered essential for devising checks and balances in which human rights are respected.

By offering this concise list of ten standards, we intend to provide practical guidance for those who seek further input for discussions, policymaking and the review of existing legislation. These standards are based on our analysis and interpretation of relevant jurisprudence, literature and selected policy documents.

Standard 1: Intelligence services need to be subject to oversight that is complete.

This means it should be complete in terms of a) the oversight body: the government, parliament, the judiciary, and a specialised (non-parliamentary, independent) commission should all play a role in oversight; b) the moment of oversight: prior oversight, ongoing oversight, and after-the-fact oversight, and c) the mandate of oversight bodies: reviews of lawfulness and effectiveness.

Standard 2: Oversight should encompass all stages of the intelligence cycle.

Surveillance involves different stages, including the collection, storage, selection and analysis of data. As all these stages amount to an interference with the right to privacy, these separate stages should be subject to oversight.

Standard 3: Oversight of the intelligence services should be independent.

In this context, this means independence from the intelligence services and the government. Judicial oversight offers the best guarantees of independence. Therefore, it is preferable to involve the judiciary in the oversight on secret surveillance and data collection.

Standard 4: Oversight should take place prior to the imposition of a measure.

In the field of secret surveillance of communications, especially by means of sophisticated technologies now associated with untargeted surveillance, the risk of abuse is high, and abuse can have harmful consequences not only for individual rights but also for democratic society as a whole. Therefore, prior independent oversight on the application of surveillance and collection powers is essential.

Standard 5: Oversight bodies should be able to declare a measure unlawful and provide for redress.

Prior and ongoing oversight bodies for intelligence services should have the power to prevent or end a measure imposed by intelligence services, and oversight bodies should have the power to declare a measure unlawful after the fact and provide for redress.

Standard 6: Oversight should incorporate the adversary principle.

The ‘adversary principle’ is a basic rule of law principle. Where secrecy is necessary, this can be implemented by the appointment of a special advocate who defends the public interest (or the interest of affected individuals). As a result, some form of adversarial proceedings would be introduced without the secrecy of measures to be imposed being jeopardised.

Standard 7: Oversight bodies should have sufficient resources to perform effective oversight.

This standard includes the attribution of the necessary equipment and staff, resources in terms of information and technical expertise. This also contributes to their independence from the intelligence services and the government.

Standard 8: Intelligence services and their oversight bodies should provide layered transparency.

This means that: a) the individual concerned, the oversight bodies, and civil society are informed; b) there is an adequate level of openness about intelligence activities prior to, during and after the fact; and c) notification, aggregate statistics, working methods, classified and detailed information about operations, and general information about what will remain secret under all circumstances is provided.

Standard 9: Oversight bodies, civil society and individuals should be able to receive and access information about surveillance.

This standard more or less mirrors the previous one. Clear legislation on receiving and access to information about surveillance must provide a framework for oversight and supports public scrutiny of the surveillance powers.

Standard 10: Companies and other private legal entities should be able to publish aggregate information on surveillance orders they receive.

Organisations should be able to disclose aggregate information publicly about orders they receive directing them to provide information to the government. They should be able to make more detailed/confidential information available to oversight bodies.

Hopefully the arguments brought forward will have a positive impact on the further development of the Dutch intelligence bill, as well as be of use to debate about intelligence legislation in other (EU) countries.

Also see the coverage by Nu.nl (in Dutch) and the coverage by Volkskrant.nl (in Dutch).

EOF

[Dutch] Bezwaar Hoofd BID dd 13 april 1971 tegen toepasbaar verklaren EVRM op IVD-en (bron: Kluiters, supplement, 1995)

Het onderstaande fragment uit ‘Kluiters deel 2’ (1995, Sdu Uitgeverij, ISBN 9012081793; p.47) citeert het Hoofd van de Buitenlandse Inlichtingendienst (HBID), die bij brief dd 13 april 1971 op onomwonden wijze bezwaar maakte tegen het, in het ontwerp-Koninklijk Besluit inzake IVD-en, expliciet van toepassing verklaren van het EVRM op de Nederlandse inlichtingen- en veiligheidsdiensten — die immers ook offensieve activiteiten ontplooien (let wel: dit speelde in het midden van de Koude Oorlog):

Op 13 april 1971 gaf HBID Hagen schriftelijk commentaar op een ontwerp voor dit KB [Voetnoot 103: CAD, MvD, ZG, map 389.834/Q. Brief dd 13-4-1971 van Hagen aan Voorzitter van de Commissie Ontwerp-KB Inl- en Veil.diensten, vadm bd H. Bos, no 02604/H/71. […]]. Om principiële en praktische redenen maakte hij bezwaar tegen de toevoeging van een artikel waarin werd gesproken over ‘gebondenheid aan de wet en het Europese Verdrag ter bescherming van enz’. Hij achtte het hoogst merkwaardig dat Nederland in een openbaar Koninklijk besluit nog eens zou verklaren dat de reeds voor het internationaal forum getekende verdragen óók van toepassing waren op de eigen staatsinstellingen. En verder: ‘De aard van het (althans offensief) inlichtingenwerk brengt juist met zich mede, dat men zich niet al te zeer aan de wet bindt en aan de wijze, waarop de rechten van de mens en de fundamentele vrijheden worden gediend, een eigen in formele zin wellicht minder toelaatbare) interpretatie durft te geven. Het offensief inlichtingenwerk zoekt en exploiteert bij de mens zwakke plekken – zij het niet in physieke, doch in psychische zin.

In deze tijd waarin men de zaken bij hun naam wil noemen moet erkend worden, dat koude oorlog óók oorlog is en dat het inlichtingenwerk nu juist niet bij uitstek uit is op eerbiediging van menselijke rechten en waardigheden.

In dit licht bezien zouden de inlichtingen en veiligheidsdiensten eigenlijk een openlijke “dispensatie” moeten ontvangen. Zover behoeft men uiteraard niet te gaan, doch de voorgestelde preambule is toch wel een ander uiterste. […] Het werk van de IDB [Inlichtingendienst Buitenland, mrk] brengt voorts met zich mede, dat de verklaring regelmatig en/of stelselmatig wordt geschonden.’

In het openbare Koninklijk besluit van 5 augustus 1972 werd uiteindelijk gesproken over ‘in gebondenheid aan de wettelijke bepalingen en in ondergeschiktheid aan Onze betrokken Minister’ [Voetnoot 104: Besluit van 5-8-1972, artikel I, 2. Staatsblad 1972, no 437].

EOF

Dutch intel oversight committee (still) seeks to publish statistics on use of special powers, suggests topics for debate on the new bill

(This is an addendum to a lengthy post on the new Dutch intelligence bill.)

In December 2014, the Dutch Review Committee on the Intelligence & Security Services (CTIVD), the independent expert committee that oversees lawfulness of Dutch intelligence practices, decided to publish statistics on the use of specific interception and on the use of the power to select from bulk-intercepted ether communication — but got censored by the Minister. While the CTIVD operates independently (it has its own million-euro budget, has full access to information, buildings and personnel, and can decide for itself what it wants to review), its reports are published via the Minister. The CTIVD can make recommendations, but those recommendations are not legally binding.

Other countries do publish some form of statistics, including Belgium (the Belgian ‘Standing Committee I’ publishes very detailed annual statistics for both VSSE and ADIV — see here), Germany, and the UK (the UK Interception of Communications Commissioner publishes aggregate statistics on interception warrants issued to MI5+MI6+GCHQ+MoD).

The CTIVD was, and still is, not amused about the Minister’s decision to censor. In June 2015, the CTIVD in its annual report upheld its opinion that statistics about the use of special intelligence powers should be openly published — notwithstanding the fact that in October 2014, in a court case brought forward by Dutch investigatory journalists, a Dutch court ruled in favor of the governments arguments to keep interception statistics secret. Here is what the CTIVD says about this on page 32 of its annual report (original translation):

The transparency pursued by the Committee in the review year was not achieved without a struggle. Publication of its findings was often preceded by discussions with the intelligence and security services and the responsible ministers, respectively. At these meetings the parties frequently achieved a satisfactory balance, but not always. Under current law the minister concerned has final say by virtue of his responsibility for national security.

In the review year, for instance, the Committee wanted to publish with respect to how many persons and organisations GISS [=AIVD] had exercised the power to intercept, in order to give society some understanding of the scale of these privacy-infringing activities. The minister of the Interior and Kingdom Relations, however, blacked out these figures in the Committee’s review report and thus made them illegible. The minister held that these figures provided insight into the current method used by GISS and must therefore be classified state secret. The Committee did not and still does not agree. The figures give an idea of the scale at which these special powers were exercised, while the outside world cannot deduce from them against which (categories of) persons and organisations the power was actually exercised. Moreover, publishing figures happens in neighbouring countries on an annual basis.

Another point of debate is how much can be disclosed to persons who complain about conduct of GISS or DISS [=MIVD]. Situations are conceivable (and have occurred) in individual cases where the interest of public accountability and awareness must take precedence over the regular policy of secrecy. The Committee discussed this above in §4.3.

An overstrong culture of secrecy not only creates scope for unacceptable practices, it may also give rise to myths and misunderstandings. As Snowden’s revelations have shown, this may eventually come to work against the intelligence and security services themselves. The Committee will continue its efforts to achieve a good balance between openness and secrecy.

In other words, it is imperative that the public debate on the new intelligence bill (details here) addresses statistics transparency as well. As far as I’m concerned, the new Dutch law should by law require statistical reports — ideally following the Belgian example. In Belgium, the oversight committee has the legal task to report statistics of special powers, but still has the legal possibility to withhold data if necessary to protect national security. The latter does however not seem to ever occur: the committee annually publishes detailed statistics. (Note: unlike the Dutch oversight committee, the Belgian oversight committee is also tasked with overseeing efficacy; I don’t know whether that may be a relevant factor here.)

The CTIVD also indicates how it intends to contribute to the public consultation of the new intelligence bill, and suggests topics for debate:

During the internet consultation process the Committee will contribute its comments on the concrete legislative texts and explanatory notes. In preparation for these comments and having regard to all the aforementioned publications and international developments in the field of intelligence and security services and their oversight, the Committee raises the following questions and issues.

1. Is the house now sufficiently in order?

In the course of its investigations the CTIVD has noticed that GISS and DISS are very much aware of the importance of privacy protection. In this sense the house is reasonably in order. There has been no systematic collection of data in disregard of the law. Nevertheless, the quality of the substantiation of the need to exercise special powers and of the reporting on such exercise is a recurring cause for concern. In fact, under the current ISS Act 2002 [=Wiv2002, the present law] the services have not yet been able to establish a procedure that ensures their consistent compliance with the statutory safeguards when selecting from untargeted interception (sigint). The Committee therefore wonders how the government thinks the services can achieve such compliance in the case of their having new and wider powers.

2. To what extent is increasing the interception powers effective and necessary?

The Committee considers it a shortcoming that up to the present there has been almost no debate on the necessity of increasing the powers of the services. The main focus of the debate is placed too readily on the lawfulness of the acts of the services and less on the efficiency or effectiveness of the interception powers. The discussion may thus never go beyond the finding that nowadays 90 percent of communications goes via the cable and that therefore the ‘traditional’ power of untargeted interception of satellite communications (the remaining ten percent) is no longer enough. But can this finding alone and by itself carry the conclusion that the powers of the services must be increased? Is it not necessary, before one can come to this conclusion, to have a picture of the effectiveness and/or the lack thereof of the existing powers? On the international level, too, this is a question which continues to be a matter of concern, without however eliciting any definite answers. The starting point should be that it must first be convincingly demonstrated that new powers are necessary because the present powers are insufficient before considering an increase of the statutory powers. The test of effectiveness also finds support in the test of legitimacy which article 8 of the European Convention on Human Rights prescribes for reasons of privacy protection. This test must not only assess the damage to national security that will be prevented, but also the harm that the powers of interception will cause to individual persons.

3. How can the privacy of innocent citizens be protected as much as possible?

The government wishes to increase the powers of untargeted interception. This means that the services will on a larger scale intercept communications of persons who are not targets of the intelligence and security services. This calls for additional obligations and safeguards. In spite of being untargeted, the interception should be ‘targeted’ as much as is possible. The data should be filtered right from the first phase of interception. The separation of relevant and non-relevant communications should be made as soon as possible after interception. Storage periods of non-relevant communications must be short and must be specifically laid down by law. Destruction of such communications should mean that the data is really and definitely destroyed. And access and use of the intercepted data must be made subject to conditions and restricted by both organisational and technical means.

4. What are the minimum requirements that must apply to the oversight of the (increased) powers of interception?

The Dessens evaluation committee makes the increase of powers conditional on reinforced oversight. It recommends in particular that the Committee’s findings of lawfulness or unlawfulness must be given binding force. The government is explicitly not following this recommendation. Notably, it puts its faith in broadening the scope of the current requirements that the ministers responsible for the performance of their tasks by the services must themselves grant permission for interception, and not in strengthening independent assessment of applications. The position and powers of the Committee are strengthened only in the field of complaints handling. International judgments appear to indicate, however, that this does not suffice to meet human rights standards in the area of privacy protection. In order to settle the issue the Committee has commissioned Leiden University to conduct a scientific study of the minimum requirements set by international law on oversight in this field. The results of the study will be published on the Committee’s website in May 2015.

EOF

Semi-public TSCM repository by James M. Atkinson (US citizenship + static IPv6 address required)

TSCM expert James M. Atkinson (more) made a large TSCM repository available via a restricted directory. To obtain access, US citizenship and a static IPv6 address are required. He announced his initiative via the following message posted to the TSCM-L mailinglist:

Many of you know that I am a pretty heavily published computer programmer, and that i am skilled in C, C++, Objective C, assembler, and embedded controllers, and programs to control devices, with a special emphasis on software that is use to control TSCM gear, and which studies the noise floor and then which detects and catalogs anything that is not noise. I have also written a great deal of software to break ciphers and codes, and published some fairly secure mechanisms of encryption that subverts even the most aggressive of eavesdropping by government funded entities.

For these who are interested, I have uploaded several tens of thousands of pages of documents in PDF format to the restricted directory on my website (www.tscm.com). The directory also contains hundreds of thousands of lines of SOURCE code for computer programs that are of value in the performance of TSCM, such as programs that you can load directly on your spectrum analyzer and have it crunch through the spectrum, and controllers I have written for all kinds of Watkins-Johnson and Ma-com gear over the decades.

There is also a large flood of ITC student texts, and texts form the National Crypto school which were acquired through a FOIA request. A huge library of .gov TSCM reports spanning decades.

In order for you to access the directory I have to issue you a user name and a password, and this user and passwords are solely for your use, on your account. Do not share that user name or the passwords as the account will automatically lock when it detects multiple people using the same account. I also need a static IPV6 address for you (if you are a TSCM person, I do not need to explain what this is) as the account is locked to your device IPV6 address.

Due to the nature of the material in this directory I will only issue User Name/Password data to a U.S. Citizen only, and the only way do to this is to present an unexpired U.S. Passport that lists you as a U.S. Citizen. Also, you have to agree in writing to not disclose the data to any other person unless the person presents to you proof that they in turn are a U.S. Citizen and the only mechanism is by them presenting an unexpired U.S. Passport to you. No un-expired U.S. passport, means zero access. Additionally, you can not have access to the directory (I will not issue a user name or password) to any person who is a convicted felon, or who has been dishonorably discharged from the U.S. Military. I also will not issue user names or passwords to any person who I have good reason to believe will disclose the information in the directory to people who can not lawfully have access to this materials. I also will not provide access to well know con artists who have a proven track record of theft of intellectual property and claiming as their own. You also actually have to be someone who actually performs big sweeps, and not merely a fringe player, or someone with spy-shop grade gear.

The first step is the passport, no passport, no access, period. Prove that you are a U.S. Citizen first in order to obtain consideration for access, then prove you are not a convicted felon or that you have been dishonorably discharged. Then ensure that your device has a IPv6 address that can be routed through your ISP.

The directory currently holds over 285,000 printed pages.

Please remember that all of the source code is of my own creation, and that I hold full copyright on it, so if you need to use it for a project you are developing, then we will need to work out suitable usage/licensing arrangements. You will notice that the software spans form 1981 to the current date, and that I am sharing the actual source code for the iPhone encryption program I developed two years ago to place a massively powerful encryption system that is fairly bullet proof into iOS.

Each user will be bandwidth restricted to 1 TB, per month in order to moderate traffic to a reasonable level.

EOF

Dutch intel bill proposes non-specific (‘bulk’) interception powers for “any form of telecom or data transfer”, incl. domestic, plus required cooperation from “providers of communication services”

UPDATE 2015-11-28: further updates moved to the bottom.

UPDATE 2015-07-09: addendum: Dutch intel oversight committee (still) seeks to publish statistics on use of special powers, suggests topics for debate on the new bill.

On July 2nd 2015, the Dutch government released (in Dutch) for public consultation the long-awaited bill (.pdf, in Dutch) + explanatory Memorandum of Understanding (MoU) (.pdf, in Dutch) that overhauls the Dutch Intelligence & Security Act of 2002 (aka “Wiv2002”). The bill is a complete rewrite of the current law (.pdf, in English), and includes expansions of power, as well as improvements to oversight, and new provisions for activities that the current law didn’t foresee (e.g. metadata analysis, small-scale use of DNA). Public consultation closes on September 1st 2015.

The post below covers, among others, the parts of the bill related to hacking and non-specific (‘bulk’) interception. First, some background.

The Wiv2002 covers both the General Intelligence & Security Service (AIVD) and the Military Intelligence & Security Service (MIVD). The cyber & sigint tasks are carried out by the Joint Sigint Cyber Unit (JSCU) that launched in June 2014. The JSCU is tasked with the collection of data from technical sources, making it accessible and searchable, perform analysis (correlation, data mining), and delivering sigint and cyber capability in support of the intelligence requirements of the AIVD and MIVD (possibly on-site in military mission areas). The JSCU will have some 350 employees. The Dutch Ministry of Defense established a Defense Cyber Command (DCC), which is affected by this bill to the extent that the DCC has relations with the MIVD or JSCU as part of its (military) tasking and operations. Roughly put, the DCC is the Dutch equivalent to USCYBERCOM, and the JSCU is the Dutch equivalent to the NSA.

Some background on the overall new interception framework — which is now divided in the phases collection, (pre)processing and analysis — is available here (note: that post was written on the basis of preliminary documents released by the government prior to release of the bill).

The approval requests that the services must send to the Minister prior to the use of most special powers are a primary source of information for legal oversight by the Dutch Review Committee on the Intelligence & Security Services (CTIVD). The CTIVD is an independent expert committee (not a court) that carries out its task on an ex post basis. For certain uses of special powers, the bill requires that the CTIVD be immediately informed by the intelligence service, and the CTIVD will then carry out an ad hoc review. If the CTIVD finds the approved practice to be illegal, the CTIVD can require the Minister to reconsider his/her decision; and then, if still necessary, inform the parliamentary standing committee on the intelligence services. The CTIVD has unrestricted access to information up to the highest classification (Dutch: “Stg. Zeer Geheim”, comparable to NATO Cosmic Top Secret), as well as to the AIVD’s building and employees. Regarding interception and hacking, neither the current law nor the bill include ex ante oversight or involvement of a court.

The non-specific interception power includes mandatory cooperation from (to be selected categories of) “providers of communication services” — a term that is defined in a way that includes not only providers of public electronic communications networks and services, but also providers of closed networks, and includes telcos, access providers, hosting providers and website operators. The use of this power requires approval from the Minister, and requires specification of the investigation (which can be long-running; think of non-proliferation and terrorism), the purpose of interception — “purpose-orientation” (Dutch: “doelgerichtheid”) is introduced as a new requirement that intends to limit bulk interception to what is relevant to a “purpose” that must be specified ‘as specific as possible’; a general indication does not suffice —, the type of telecommunications (e.g. GSM, radio, satellite, internet; optionally including geographic boundaries), optionally the types of traffic that are relevant (e.g. voice, chat, file transfer), and in the case of cable networks, the cable infrastructure that is targeted. In other words, no blanket authorizations for non-specific interception will exist, although blanket-like authorizations may, depending on how broad a “purpose”, in the context of a specified investigation, is allowed to be in practice; the requirement “as specific as possible”, mentioned in the MoU, leaves room for interpretation.

Specific interception, i.e., interception that only targets a specific individual, organization or technical characteristic (phone number, IP address, etc.), is omitted from this post, as that part of the law is not notably changed. The hacking power is included because it includes a new paragraph aimed at reconnaissance of computer networks (e.g. mapping computers and networks, running port scans, inquiring software/hardware configurations, etc.; think of GCHQ’s HACIENDA).

Now, moving on to selected texts from the bill and the MoU.

WARNING: the below contains unofficial translations. Feel free to contact me if you have questions.

The hacking power is pretty much unchanged, except that a specific provision is introduced for reconnaissance (Art. 30-1-a):

Article 30

  1. The services are authorized to:
    1. explore the technical characteristics of automated works that are connected to a communications network;
    2. whether or not using technical interference, false signals, false keys, false identity or through intervention of the automated work of a third party, access an automated work;
  2. The power meant in the first paragraph, under b, includes the power to:
    1. break any security;
    2. make technical provisions to undo the encryption of data stored or processed in the automated work;
    3. make technical provisions in relation to exercising the power meant in Article 25, first paragraph [=observation of persons], and Article 32, first paragraph [=specific interception];
    4. take over the data stored or processed in the automated work.

[…]

The MoU mentions that the power meant in the first paragraph, under a, will be used “semi-continuously” in order to detect relevant changes and keep an up-to-date picture of parts of digital infrastructure relevant to specific investigations. The MoU explains that the hacking power also includes the possibility of hacking a third-party system if a target’s own system cannot be directly broken into (as seen in Art.30-1-b):

The technical reality shows that targets are generally security-aware, but that operational opportunities for using weaknesses in technical peripheral users, such as co-tenants of a certain server, which can lead to successful breaking into the automated work of the target.

(Side note: to non-targets, some reassurance might be found in the following remarks in CTIVD oversight report 19 (.pdf, 2009) considering the interception of communication of non-targets, assuming its findings still apply today:

The AIVD can exercise a special power against a person in the (immediate) environment of the target, in order the obtain information about the target via this person (the non-target). The Committee considers this to be an extremely heavy means and finds that the AIVD must be very reluctant with its use. This means in not applied in large numbers and in nearly all cases in which the means is applied, the Committee finds that it is necessary as result of a threat to national security, because other means do provide insufficient insight into the target.)

The bill then introduces the term “provider of a communication service”, derived from the term “service provider” introduced in the Budapest Convention on Cybercrime (2001) (note: possibly, the official English translation of the Dutch law will mention “service provider”; I translated the text as literal as possible):

Article 31

In this paragraph and the provisions based thereon, the following definitions apply:

  1. provider of a communication service: the natural or legal person who, in carrying out their profession or business, offers users of the service the possibility to communicate via an automated work, or who processes or stores data for the purpose of such a service, or for a user of that service;
  2. user: the natural or legal person who has a contract with the provider of a communication service concerning the use of that service or who actually uses such a service.

[…]

From the MoU it is clear that “provider of communication service” at least includes providers of public telecommunication services and networks (public telcos and internet access providers) and providers of closed services and networks, as well as hosting providers (e.g. cloud) and website operators. The paragraphs cited below will apply to a “limited number” of such entities, in categories that will be determined by governmental decree.

(Side note: the term “automated work” is linked to the Dutch computer crime legislation. The Dutch government is preparing another bill that will grant police hacking powers. That bill won’t be released until after the parliamentary summer break, which ends on August 31st 2015. Some details available here.)

The intelligence services are granted the power of non-specific interception of “any form of telecommunications or data transfer via an automated work” (cable, ether, whatever; regardless of its source & destination, thus including domestic communication):

Article 33

  1. The services are authorized to, using a technical aid, wiretap, receive, record and listen to any form of telecommunications or data transfer via an automated work regardless of location in other cases than meant in Article 32 [= the specific interception power], if what has been required or provisioned by law is complied with. The power, meant in the previous sentence, includes the authority to undo encryption of telecommunications or data, as well as technical analysis of the data, insofar this is aimed at optimizing the use of the power meant in the previous sentence.

[…]

Article 34 provides the power to carry out sigint search/exploration on bulk-intercepted data (e.g. DPI, analyzing traffic flows, anomaly & signature based intrusion detection, etc.):

Article 34

  1. The services are authorized to carry out research using the data collected on the basis of the power meant in Article 33, for the purpose of:

    1. determining the characteristics and nature of the telecommunication;
    2. determining the identity of the person or organization related to the telecommunication.
  2. The services are furthermore, in the context of exercising the power meant in Article 35, second paragraph, authorized to carry our research on the basis of data collected on the basis on Article 33, for the purpose of:

    1. determining and verifying selection criteria related to persons and organizations or keywords related to topics of investigations;
    2. identifying persons or organizations, who in the context of ongoing investigations are eligible for being investigated by the service.

[…]

Note that the current sigint search power (Article 26 Wiv2002) is limited to communication that has a foreign source and/or destination. The above article in the bill no longer contains that restriction, thus enabling sigint search on domestic communications as well.

Providers of communication services can be required to hand over data that is needed to exercise the non-specific interception power (exercise of this power does not require separate approval from the Minister):

Article 36

  1. The services are authorized to request a provider of a communication service to provide data, which are necessary to exercise the power meant in Article 33, first paragraph. The categories of data, to which the request mention in the previous sentence can apply, will be determined by governmental decree.

[…]

  1. The provider of a communication service is required to comply with a request as meant in the first sentence of the first paragraph.

[…]

The MoU sheds a dim light on what data is meant in Article 36-1 (p.79):

[…] This involves acquiring information that can help map the communications landscape, which is necessary to, at some point in time, exercise the interception power meant in Article 33.[…]

[…] This involves, among others, the technical data of for instance the telecommunications network exploited by the provider, and the equipment used etc., which are necessary to — in consultation with the provider — determine what technical provisions that need to be made to carry out the authorized interception. […]

Footnote 63 of the MoU sheds slightly more light on this, and refers to the newly introduced requirement that interception must be purpose-oriented:

63: In order to intercept in a purpose-oriented manner, it must be made clear where, what type of communication is processed c.q. transported. This involves for instance information concerning business customers/tenants and data commonly known as part of daily operations of providers of communication services about the services offered, characteristics of traffic flows, and communication channels.

The data meant in Article 36 hence include data about the physical and/or logical layout of infrastructure, routing, signal properties, etc.

Providers of communication services must, at the request of the intelligence services as approved by the Minister, cooperate in enabling the intelligence services to exercise their non-specific interception powers, in that the providers must provide access to their systems/networks:

Article 37

  1. The services are authorized to request a provider of a communication service to cooperate in exercising the authorized interception as meant in Article 33, second paragraph.

[…]

  1. The provider of a communication service that is not already required to cooperate based on Article 13.2 of the Telecommunications Act, is required to comply with a request as meant in the first paragraph. The services are authorized to contact a provider of a communication service to request cooperation in the exercise of an authorized request as meant in Article 33, second paragraph.
  2. The provider must maintain, for twelve months, the technical provisions made as part of the requested cooperation as authorized per the second paragraph [= Ministerial authorization].

[…]

Per Article 132, not complying with an order is punishable as felony (if intent can be proven) or misdemeanor (if intent can not be proven). Page 202 of the MoU indicates that use of Article 37 will initially be limited to “several physical access points” (at telcos and/or internet exchanges):

(…) Partially to gain experience, on the basis of which more specific next steps can be taken, the interception will be limited to several physical access points in the first years after the law is enacted.

Raw bulk intercepts can now remain available for selection (by persons, organizations, technical characteristics and topic for keyword-based searches) three years instead of one year, as is presently the case. This increase is due to problems the services experience concerning (long-running) investigations into non-proliferation. Article 49 allows the services to share raw bulk intercepts — metadata and contents — with foreign intelligence & security services, under certain conditions, and only if approved by the Minister.

Providers of communication services can also be required, after approval from the Minister, to hand over users’ (stored) telecommunications contents (Art. 38; e.g. this is not a bulk power, telecommunications data can only be requested for a specified person, number, account, etc., such as the mailbox(es) of specified webmail users):

Article 38

  1. The services are authorized to contact a provider of a communication service to request data concerning the telecommunication of a user that has been stored by the provider as part of the communication service offered. The categories of data, to which the request mention in the previous sentence can apply, will be determined by governmental decree.

[…]

  1. The provider of a communication service is required to comply with a request as meant in the first sentence of the first paragraph.

[…]

Furthermore, the intelligence services are authorized, under certain conditions and after approval from their Minister (Art.30-6 and Art.41-2), compel anyone (Dutch: “een ieder”) — also including organizations — to help decrypt data in an automated work (Art.30-5 to 30-8) or help decrypt conversations, telecommunications or data transfer (Art.41-1), e.g. by handing over keys or providing decrypted data. (A similar provision is present in the current law.) Another legal option to defeat encryption is the use of the hacking power (Art.30, see below), which requires after approval from the Minister; and yet another legal option is the use of agents (who can be tasked with interception or hacking) or informants (e.g. a sysop who, as part of daily work, has access to cryptographic keys). Again, per Article 132, not complying with such an order is punishable as felony or misdemeanor.

New provisions are present concerning “automated data analysis” — think of metadata analysis based on non-specific intercepts:

Article 35

  1. The services are authorized to:
    1. select the data that have been collected through the use of the power meant in Article 33.
    2. apply automated data analysis as meant in Article 47 concerning data collected using Article 33 that concerns data other than the content of that telecommunication.

[…]

Article 47

  1. The services are authorized to apply automated data analysis concerning:
    1. data from the services’ own automated databases,
    2. data from information sources accessible to anyone,
    3. data from automated databases to which the services have direct automated access, and
    4. data from databases provided by third parties.
  2. For the purpose of processing the data meant in the first paragraph the data can at least:
    1. be compared in an automated way, or be compared in combination with each other;
    2. be searched on the basis of profiles;
    3. be compared for the purpose of tracing certain patterns.

[…]

In Article 47-2 the words “can at least” mean that the list (a, b, c) in Article 47-2 is not exhaustive. Concerning the profiling ex Art.47-2-b, the MoD states that hits found during automated profile matching must first be analyzed by a human before measures can be taken against the person(s) that match a profile.  The “data from databases provided by third parties” (Art. 47-1-d) refers to databases that are provided voluntarily ex Article 22; there is no power to compel third parties (e.g. the private sector) to provide data(bases) in the way that US intelligence services can under FISA Section 215 on the basis of the “tangible things”-provision.

 

Lastly, Article 28 introduces provisions for the processing and storing of DNA by the intelligence services. This follows recommendations made in an oversight report concerning the lawfulness of the services’ (small-scale) use of DNA; some details available here.

The bill still restricts the use of the interception (non-specific and specific) and hacking powers to specific legal tasks, but a new task has been added for both the AIVD (‘g-task’) and MIVD (‘h-task’) concerning security screening of agents and informants (but excluding security screenings as meant in the Security Screenings Act (Dutch: “Wet veiligheidsonderzoeken”, or “Wvo”), e.g., the screening of candidate-employees of the intelligence services, defense industry, etc.). Interception is currently not permitted in that context. For the AIVD, use of special powers remains restricted to their security task (‘a-task’; think of national security) and their foreign intelligence task (‘d-task’; think of non-proliferation). For the MIVD, use of special powers remains restricted to their task concerning enemy forces (‘a-task’), their task concerning the Dutch armed forces  (‘c-task’) and their military foreign intelligence task (‘e-task’). According to oversight report 38 (in Dutch), raw data from non-specific interception can not be used for the services’ other legal tasks, but evaluated data (i.e., data that has been processed and analyzed) can.

Article 18 limits — as does current law — processing of personal data by the services to specific categories of relevant persons (targets, persons voluntarily undergoing a security screening, etc.). But as the MoU explains, automated data analysis (e.g. metadata analysis) can involve processing data, e.g. collected via bulk intercepts, from other persons (non-targets) as well. And recall that raw bulk intercepts can be retained for three years.

Article 22 provides — as does current law (details) — the services permission to ask private parties (e.g. banks, public transport, or candidate-informants elsewhere) to voluntarily hand over data or databases, or make those accessible via automated means. The handover of personal data is exempted from Dutch Data Protection Act (“Wbp”). Article 22 combined with Article 35-1-b and/or 47 enables profiling, social network analysis etc. on the basis of voluntarily provided data; in addition to possible combination with data collected through other means, such as bulk interception, specific interception, and hacking. The related provision in the current law is used, for instance, in the context of tracking terrorist’s finances. Requests for data can be made only in the context of a specified purpose/investigation; the MoU states: “arbitrary requests for data are not allowed”.

Someone told me the bill seems quite polarizing; indeed, there are a lot of “musts” present, and the definition of “provider of a communication service” is very broad. But further things are relevant to assessing the bill as a whole:

  • the proposed (permanent) five-year mandatory re-examination of the law (Article 147);
  • the proposed introduction of required “purpose-orientation”, which intends to limit the hay stack created using non-specific interception to relevant information. The MoU mentions that “filters” will be used to filter (ir)relevant data;
    • caveat: it remains to be seen what filters will be applied (filters for volume reduction? of course; filters for privacy? only if possible) and how broad a “purpose” can be defined for the collection phase and the (pre)processing phase in the new interception framework;
  • the proposed provisions that provide new ways of reporting wrongdoing (Article 114-120). The provisions extend to ‘anyone involved in the exercise of this law’ (including e.g. employees of providers) and cover reporting violations of law, dangers to security, and dangers to the proper functioning of “the public service” (the latter is mentioned in Article 114-c but not explained in the MoU; presumably it refers to the intelligence service);
  • the proposed mandatory reporting about the use of special powers (Article 45);
    • caveat: related existing requirements have not always been met in the past;
  • the proposed limitations on retention of intercepts — raw bulk intercepts can be stored for three years ex Art. 33-5, irrelevant yields of hacking, specific interception and obligated telecoms data hand-over can be stored for max 12 months ex Art. 30-9, 32-10 and 38-7;
    • caveat: encrypted data obtained via non-specific interception can (still) be stored indefinitely; the retention period does not start until the data is decrypted — think of collecting TLS-encrypted traffic until you know what to do with it (possibly compel someone to decrypt it or hand over keys). Also, raw bulk intercepts can now remain available for selection three years instead of one year, as is presently the case;
  • the proposed specification of information that must be present in approval requests sent to the Minister. Concerning bulk interception, the MoU states that the approval request must specify the relevant ongoing investigation, the purpose of the requested interception (that must be specified ‘as specifically as possible’; ‘a general indication does not suffice’), the type of telecommunications (e.g. GSM, radio, satellite, internet; ‘possibly’ including geographic boundaries), ‘possibly’ the types of traffic that are relevant (e.g. voice, chat, file transfer), and, in the case of bulk cable interception, the cable infrastructure that is targeted.
    • caveat: it has always been required that approval requests include a motivation (in terms of necessity, proportionality and subsidiarity), but oversight reports show that it has often been missing or was inadequate in the case of the use of the existing (ether-only) sigint power — which is why I stated oversight is currently broken. The new bill aims to fix that (e.g. through the new three-phase interception framework, and by no longer requiring separate approval requests from the Minister prior to selecting persons, organizations, characteristics or keywords from raw bulk intercepts), but we won’t really know what has (not) been fixed until the bill is adopted and new oversight reports are published based on the new legal framework;
  • the proposed increase of the level of authorization required for hacking (ex Art.30-3) and sigint search (ex Art.34) from head of service to Minister;
    • caveat: the risk of rubber-stamping remains a point of attention, because — as is clear from past oversight reports and parliamentary papers — the Minister usually has a lot of approval requests to decide on. In fact, the number of requests is likely to increase as result of the heightened approval regime for various (old and new) powers; although sigint selection (selecting data about specific persons, organizations, technical characteristics, or keywords from raw bulk intercepts) no longer requires separate, per-case approval from the Minister;
  • the many other proposed changes concerning oversight — but alas, still no ex ante or court oversight —, complaint handling, aspects of freedom of information, and more.

The compelled-decryption provision laid down in Article 41 references, in Article 41-1, the non-specific (‘bulk’) interception power laid down in Article 33-1. If I’m interpreting the bill and MoU correctly, this permits the Minister to compel organizations (such as providers) to decrypt data or hand over keys also for the purpose of exercising the non-specific (‘bulk’) interception power. Think of requiring handover of TLS keys used on shared servers, cloud CDNs, or crypto used to protect links between data centers. If this is indeed in scope of the law, then that’s probably a topic for debate — regardless of the fact that page 202 of the MoU indicates that use of the compelled-access power will be limited to “several physical access points” in the initial years, and without leaving this up to assumptions about whether the intelligence services will or will not apply the law this way, and whether the Minister and CTIVD would approve in specific cases.

Appendix 3 of the MoU provides an overview of all powers and safeguards in the bill:

20150702_wiv-voorstel-bijlage3-en

(Also available as .pdf here.)

UPDATES (new to old)

UPDATE 2016-04-29: the confidential revised draft + MoU that the Dutch government submitted for advice to the Council of State was leaked via Dutch newspaper Volkskrant. An analysis by Bits of Freedom is available here (in English). I posted some notes on my blog here and here (both in Dutch).

UPDATE 2016-04-15: an additional NOS report states that the existing (targeted) interception power and (targeted) hacking power, too, will be subject to ex ante and binding oversight. The newly to be established oversight committee is entitled ‘Toetsingscommissie Inzet Bevoegdheden’ (TIB), and will consist of persons that have a background in the judicial branch.

UPDATE 2016-04-14: NOS reports that according to unnamed sources, the Dutch government will, with regard to the upcoming cable interception power, consider ex ante oversight by a new independent committee. (Note: ‘new independent committee’ presumably means it’s not about the CTIVD, the existing independent expert committee that has been carrying out ex post non-binding oversight since 2002.) Furthermore, requiring prior court approval for the interception of communications of lawyers and journalists will also be considered. The report states these topics will be discussed in the cabinet. The cabinet changed “purpose-oriented interception” (my translation from the Dutch word “doelgericht” used previously) to “investigation-bound interception” (my translation of “onderzoeksopdrachtgericht”). It’s just different words for the same thing: interception that is carried out in the context of an intelligence task, but that is not targeted/limited to specific known persons or organizations.

UPDATE 2015-11-11: the transcription of a parliamentary debate of September 2nd 2015 on intelligence affairs, including the draft bill, is now available here (in Dutch). When the cabinet submits the revised bill to the House of Representatives (lower house), the house will also receive the Privacy Impact Assessment (PIA) performed on the draft bill by PI.lab (a cooperation between Radboud University, Tilburg University, SIDN and TNO) and coordinated by professor Bert-Jaap Koops.

UPDATE 2015-11-04: the oversight committee (CTIVD) published its response (in Dutch) to the draft bill. An English outline of that response is available here.

UPDATE 2015-10-21: the current progress of the bill within the legislative process can be viewed at the ‘live’ webpage here (in Dutch). The bill is currently at the end of the preparatory phase: public consultation closed on September 1st, and the government is now processing the public comments. Next, the government will submit the bill, possibly changed as result of public comments, to the Council of State for advice. Then, the government will submit the bill, possibly changed as result of the Council’s advice, to the House of Representatives. If the bill is adopted by a majority vote of the House, the bill, possibly changed, will be submitted to the Senate. If it is adopted by the Senate, the law and a date of enactment will be published. Conceivably, whatever the new law will look like, it seems likely it won’t be enacted before 2017.

UPDATE 2015-07-23: today, three weeks after publication of the bill, legal scholars from the University of Amsterdam published a report (.pdf, in English) entitled “Ten standards for oversight and transparency of national intelligence services”. Short write-up here. Notably, the authors of the report plead for ex ante review of interception and intelligence operations by a court. The bill neither includes independent ex ante review, nor any court involvement.

EOF