Posts

Double Dipping at SWIFT

/16 Comments/in , /by

Spiegel today reveals more details about NSA’s “Follow the Money” program, in which it collects credit card information from select geographical regions. In addition, as TV Globo also revealed last week, they are conducting Tailored Access Operations against SWIFT, the international financial transfer messaging system.

The NSA’s Tracfin data bank also contained data from the Brussels-based Society for Worldwide Interbank Financial Telecommunication (SWIFT), a network used by thousands of banks to send transaction information securely. SWIFT was named as a “target,” according to the documents, which also show that the NSA spied on the organization on several levels, involving, among others, the agency’s “tailored access operations” division. One of the ways the agency accessed the data included reading “SWIFT printer traffic from numerous banks,” the documents show.

Now, some caution about this claim is in order. Spiegel reports that NSA’s financial records database has 180 million records, of which 84% are credit card transactions.

The collected information then flows into the NSA’s own financial databank, called “Tracfin,” which in 2011 contained 180 million records. Some 84 percent of the data is from credit card transactions.

Even assuming the balance of the records in the database come from SWIFT, that’s less than 29 million records (in 2011, so assume the number is larger now). In 2011, SWIFT was sending 17.5 million records a day. So whatever makes it into the actual database is just a small fraction of international traffic.

But that almost certainly doesn’t account for the bulk of the SWIFT information collected by the US government. Remember: in addition to stealing the data, Treasury also gets it via a now-public agreement. The former CEO of SWIFT Leonard Schrank and former Homeland Security Czar, Juan Zarate actually boasted in July, in response to the earliest Edward Snowden revelations, about how laudable Treasury’s consensual access to the data was.

The use of the data was legal, limited, targeted, overseen and audited. The program set a gold standard for how to protect the confidential data provided to the government. Treasury legally gained access to large amounts of Swift’s financial-messaging data (which is the banking equivalent of telephone metadata) and eventually explained it to the public at home and abroad.

It could remain a model for how to limit the government’s use of mass amounts of data in a world where access to information is necessary to ensure our security while also protecting privacy and civil liberties.

Never mind that by the time they wrote this, an EU audit had showed the protections were illusory, in part because the details of actual queries were oral (and therefore the queries weren’t auditable), in part because Treasury was getting bulk data. But there was a legitimate way to get data pertaining to the claimed primary threat at hand, terrorism. And now we know NSA also stole data.

Note, too, the timing. While Spiegel doesn’t provide enough details about the exploitation of SWIFT for us to date it, the dates it does provide about this financial spying are 2010 and 2011. That was the period when the EU was trying to put sensible limits to Treasury’s access of SWIFT.

Back when the intelligence community first decided to go after SWIFT data, their first plan was to just steal it.

Intelligence officials were so eager to use the Swift data that they discussed having the C.I.A. covertly gain access to the system, several officials involved in the talks said. But Treasury officials resisted, the officials said, and favored going to Swift directly.

12 years later, they apparently are stealing at least some of it. That probably means they wanted data for transactions that have nothing to do with the counterterrorism application first SWIFT and then the EU bought off on. So there’s the legal access to counterterrorism data via Treasury, and the illegal access to (presumably) some other kind of data via NSA.

Read more →

SWIFT: Big Brother with a Booz Assist, Only without the Paperwork

/14 Comments/in , /by

As reporting on Edward Snowden reveal the scope of our spying on European friends, I’ve been thinking a lot about SWIFT.

SWIFT, you recall, is the database tracking international online money transfers. After 9/11, the US Government started helping itself to the data to track terrorist financing. But then in 2010 the servers moved entirely to the EU, and the EU forced the US to accede to certain protections: protections for EU citizens, a prohibition on bulk collection (and with it data mining), and two-pronged audit system.

Today, the CEO of SWIFT until 2007, Leonard Schrank, and the former Homeland Security Advisor, Juan Zarate, boast about the controls on SWIFT, suggesting it provides a model for data collection with oversight.

Both the Treasury and Swift ensured that the constraints on the information retrieved and used by analysts were strictly enforced. Outside auditors hired by Swift confirmed the limited scope of use, and Swift’s own representatives (called “scrutineers”) had authority to stop access to the data at any time if there was a concern that the restrictions were being breached. These independent monitors worked on site at government agencies and had real-time access to the system. Every time an analyst queried the system, the scrutineer could immediately review the query. Each query had to have a reason attached to it that justified it as a counterterrorism matter. Over time, the scope of data requested and retained was reduced.

This confirmed that the information was being used in the way we said it was — to save lives.

[snip]

The use of the data was legal, limited, targeted, overseen and audited. The program set a gold standard for how to protect the confidential data provided to the government. Treasury legally gained access to large amounts of Swift’s financial-messaging data (which is the banking equivalent of telephone metadata) and eventually explained it to the public at home and abroad.

It could remain a model for how to limit the government’s use of mass amounts of data in a world where access to information is necessary to ensure our security while also protecting privacy and civil liberties.

This description should already raise concerns about the so-called gold standard for spying. When “scrutineers” cohabit with those they’re supposed to be scrutinizing, it tends to encourage cooperation, not scrutiny.

And somehow, Schrank and Zarate neglect to mention that the vaunted audit process they describe was conducted by none other than Booz Allen Hamilton, the contractor that hired and let Edward Snowden abscond with the spying world’s crown jewels. And, as ACLU noted in a report for the EU in 2006, even during Schrank’s tenure, Booz was neck deep in aggressive surveillance.

But the real problem with highlighting SWIFT as a poster child of massive surveillance done right post-dates Schrank’s tenure (though he must know about this), when the EU’s independent audits for the first time revealed what went on in SWIFT queries. Among other things: the actual requests were oral, and therefore couldn’t be audited.

The report revealed that the Americans have been submitting largely identical requests–but then supplementing them with oral requests.

The oral requests, of course, make it impossible to audit the requests.

At the time of the inspection, Europol had received our requests for SWIFT data. Those four requests are almost identical in nature and request–in abstract terms–broad types of data, also involving EU Member States’ data. Due to their abstract nature, proper verification of whether the requests are in line with the conditions of the Article 4(2) of the TFTP Agreement–on the basis of the available documentation–is impossible. The JSB considers it likely that the information in the requests could be more specific.

Information provided orally–to certain Europol staff by the US Treasury Department, with the stipulation that no written notes are made–has had an impact upon each of Europol’s decisions; however, the JSB does not know the content of that information. Therefore, where the requests lack the necessary written information to allow proper verification of compliance with Article 4(2) of the TFTP Agreement, it is impossible to check whether this deficiency is rectified by the orally provided information. [my emphasis]

In addition, in spite of demands that the program include no bulk downloads, that’s precisely what the US was doing.

“We have given our trust to the other EU institutions, but our trust has been betrayed”, said Sophia in’t Veld (ALDE, NL), rapporteur on the EU-US Passenger Name Record (PNR) agreements. “This should be kept in mind when they want our approval for other agreements”, she declared.

“Somehow I am not surprised”, said Simon Busuttil (EPP, MT), recalling that “at the time of the negotiations last year we were not satisfied with having Europol controlling it – we wanted additional safeguards”. He added that ”the agreement is not satisfactory”, since it involves the transfer of bulk data, and insisted that ”we need an EU TFTP”.

For Claude Moraes (S&D, UK), the US demands are “too general and too abstract”. He also recalled that MEPs had insisted at the time that it must be specified how the US request would be made and that they needed to be “narrowly tailored”. A written explanation should accompany each request, he added.

This agreement is not in line with Member States’ constitutional principles and with fundamental rights, argued Jan Philipp Albrecht (Greens/EFA, DE). He highlighted the problem of bulk data transfer, “which is exactly what we have criticised before“. [my emphasis]

In other words, once an actual independent reviewer — not an embedded contractor like Booz — reviewed the program, it became clear it was designed to be impossible to audit, even while engaging in precisely the bulk downloads the Europeans feared.

Not only is the experience of SWIFT one reason why the Europeans are so quick to object to the scale of US spying on them. But it is actually a poster child for surveillance done wrong.

Contrary to what its boosters want you to believe.