Posts

How a Russian Dangle about Shadow Brokers Started Dictating NSA’s Twitter Feed

As you may know, we’ve been fostering dogs. Our current dog, June Bug (pictured above), is a terrorist. She’s really smart. She creates diversions so she can try to steal our food. We can only get her to play with dog toys if we “trick” her, by hiding them in boxes that she first destroys. But today, she got outfoxed (heh) by a squirrel. We were walking south towards a bush and a big oak and she saw the squirrel under the bush. While we were walking past the oak, the squirrel bolted up the oak so high that June Bug (who at least is better at understanding a third dimension than McCaffrey the Millennial Lab was) couldn’t see her. June Bug kept looking under the bush until finally she turned to the oak but by then the squirrel was well beyond her vision up in the oak.

This story, reported in both the Intercept and the NYT, on the CIA and NSA’s efforts to reach out to Russia to get Shadow Brokers tools feels like that exchange. Reading the two in tandem, it’s clear that the Russians learned the CIA and NSA were trying to buy back the tools released by Shadow Brokers, and used the channel the US set up with a Russian “businessman” to provide likely disinformation about Trump’s ties to Russia instead. NYT describes obtaining,

Russian produced unverified and possibly fabricated information involving Mr. Trump and others, including bank records, emails and purported Russian intelligence data.

[snip]

All are purported to be Russian intelligence reports, and each focuses on associates of Mr. Trump. Carter Page, the former campaign adviser who has been the focus of F.B.I. investigators, features in one; Robert and Rebekah Mercer, the billionaire Republican donors, in another.

The Intercept said the government even obtained an FBI report that had been purloined.

Recently, the Russians have been seeking to provide documents said to be related to Trump officials and Russian meddling in the 2016 campaign, including some purloined FBI reports and banking records.

It’s equally clear that, as things soured, the source reached out to James Risen to make sure the story would come out with the spin that the CIA had cut off the exchange because it didn’t want to receive dirt on Trump. Note, the NYT story doesn’t include the agency split.

What’s perhaps most embarrassing about the story is that the NSA tweeted out pre-arranged tweets at least ten times (the Intercept describes which tweets they were) as a signal that the American businessman intermediary was really working on behalf of the US government. The last that Risen lists is one pertaining to Section 702 on December 13.

Effectively, Russia was yanking NSA’s chain, and possibly tracking communication pathways from the American intermediary through NSA to the Twitter feed.

The incident is interesting for several reasons. First, it may corroborate the “second source” theory I posited back in September (which I was pretty sure was in the neighborhood in any case given some curious attention the post got). It seems to confirm that the spooks at least came to believe that Russia was behind the Shadow Brokers and Vault 7 compromises (though Russia doesn’t appear to have shared any legitimate non-public files, so it’s not necessarily proven).

Trump is now using this effort at disinformation the same way he has used the Steele dossier: in a bid to claim his own innocence.

I’m perhaps most interested in the timing of this. The government seemed to treat the Nghia Hoang Pho plea in early December as its explanation for how the Shadow Brokers files got stolen. If that’s true, it should know what Russia or whoever else took (or they could at least ask Kaspersky nicely, which seems to have a pretty good idea of what was there). It wouldn’t need to chase this intermediary for two more months.

And yet they did.

Has Hal Martin Finally Gotten the Government to Admit He Didn’t Feed Shadow Brokers?

Hal Martin may finally get a plea deal.

On Tuesday, Martin’s (excellent) public defender James Wyda asked to cancel a guilty plea to one of the 20 charges against him which had been scheduled for next week, stating that continuing negotiations may settle the whole case.

The defense requests a cancellation of the Rule 11 guilty plea hearing currently scheduled for January 22, 2018. The parties are continuing negotiations with the hope of resolving the entire case.

As John Gerstein had previously reported, last month Martin unilaterally moved to plead guilty to retaining one document described as “a March 2014 NSA leadership briefing outlining the development and future plans for a specific NSA organization,” though the government still threatened to ask for the maximum sentence on that one charge. But something changed since then to reinvigorate plea discussions.

I’m particularly interested in the schedule Judge Marvin Garbis had set in response to Martin’s bid to plead to one charge. The plea would have triggered a CIPA review, the process by which judges decide what classified information is necessary for a criminal trial, often in substitute form.

This is to confirm, as stated at the conference held this date:

1. On January 8, 2018, Defendant shall file a letter including its version of the statement of facts as to Count One of the Indictment.

2. Defendant Martin intends to plead guilty to Count One on January 22, 2018 at 10:00 A.M.

3. Defendant Martin expects to file a CIPA § 4 submission on January 26, 2018.

4. The Government shall make an ex parte presentation regarding its contentions and its pending CIPA § 4 motion in an on-the-record sealed proceeding on February 1, 2018 commencing at 10:00 A.M.

5. Defendant Martin shall make an ex parte presentation regarding its contentions and its forthcoming CIPA § 4 submission in an on-the-record sealed proceeding at a time to be scheduled by further Order.

That’s presumably an indication that Martin wanted to use classified evidence to mitigate his sentence. And all of this has happened in a six week extension Martin’s lawyers asked for on December 8, explaining that they had only just gotten access to information seized (back in August 2016) from Martin’s car and home.

On November 28, 2017, we had the opportunity to conduct an evidence review at the Baltimore FBI Field Office’s Sensitive Compartmented Information Facility for the first time of some of the items allegedly seized from Mr. Martin’s car and residence. In light of the volume of material made available for our review, we expect to return to the FBI multiple more times to review the remainder of the items.

All of which suggests the defense saw something in their classified discovery that made them think they can mitigate Martin’s sentence and, possibly, eliminate the government’s interest in trying him for those other 19 retained documents.

So to recap: on December 8, Martin’s lawyers ask for more time. On December 22, he moves to plead guilty. In the last few weeks, the judge set in motion the process to allow Martin to use classified information in his sentencing (and his lawyers submitted their version of what he would plead guilty to). And now a plea deal may be in the offing.

All that happened in the wake of Nghia Hoang Pho pleading guilty on December 1, after some interesting timing delays as well, timing which I laid out here.

The actual plea deal is dated October 11. It states that “if this offer has not been accepted by October 25, 2017, it will be deemed withdrawn.” The information itself was actually signed on November 29. Friday, the actual plea, was December 1.

So while there’s not a substantial cooperation component in the plea deal, certainly a substantial amount of time took place in that window, enough time to cooperate.

And consider the news coverage that has happened during that period. The initial plea offer was made in the week following a big media blitz of stories blaming Pho (and through him Kaspersky) for the Russian theft of NSA tools. In the interim period between the offer and the acceptance of the plea deal, Kaspersky confirmed both verbally and then in a full incident report that his AV had found the files in question, while noting that a third party hacker had compromised Pho’s machine during the period he had TAO’s tools on it.

In other words, after at least an 18 month investigation, Pho finally signed a plea agreement as the media started blaming him for the compromise of these tools.

In that plea deal, the government noted that they could have charged Pho as they had charged Martin, with one count for each retained file (though in reality Martin got charged for a tiny fraction of what he brought home).

During much of that period, Harold Martin was in custody and under investigation for a similar crime: bringing a bunch of TAO tools home and putting them on his computer. Only, unlike Pho, Martin got slammed with a 20-count indictment, laying a range of files, and not just files from NSA. Indeed, the Pho plea notes,

This Office and the Defendant agree that the Defendant’s conduct could have been charged as multiple counts. This Office and the Defendant further agree that had the Defendant been convicted of additional counts, … those counts would not group with the count of conviction, and the final offense level would have increased by 5 levels.

That is, the government implicity threatened Pho to treat him as Martin had been, with a separate charge tied to the individual files he took.

Now, perhaps that’s all that Martin’s lawyers were going to note, that a similarly situated defendant in the same district had been able to plead guilty to a single charge.

But I wonder if there’s not more, specifically related to that plea, pertaining to the real source of the Shadow Brokers files. That is, if Pho was permitted to plead guilty after having making the Shadow Brokers files accessible to third party hackers coming in after Kaspersky’s AV got shut down, then why couldn’t Martin, whose files were air gapped from such measures, obtain a similar plea?

The Spooks Struggle with Reciprocity

I’ve written a lot about the norms (or lack thereof) that the US might set by indicting nation-state hackers for their spying. Notably, I was the first to formally note that Shadow Brokers had doxed some NSA hackers in his April release.

On Friday, along with details about previously unknown, very powerful Microsoft vulnerabilities and details on the 2013 hacking of the SWIFT financial transfer messaging system, ShadowBrokers doxed a number of NSA hackers (I won’t describe how or who it did so — that’s easy enough to find yourself). Significantly, it exposed the name of several of the guys who personally hacked EastNets SWIFT service bureau, targeting (among other things) Kuwait’s Fund for Arab Economic Development and the Palestinian al Quds bank. They also conducted reconnaissance on at least one Belgian-based EastNets employee. These are guys who — assuming they moved on from NSA into the private sector — would travel internationally as part of their job, even aside from any vacations they take overseas.

In other words, ShadowBrokers did something the Snowden releases and even WikiLeaks’ Vault 7 releases have avoided: revealing the people behind America’s state-sponsored hacking.

Significantly, in the context of the SWIFT hack, it did so in an attack where the victims (particularly our ally Kuwait and an apparent European) might have the means and the motive to demand justice. It did so for targets that the US has other, legal access to, via the Terrorist Finance Tracking Program negotiated with the EU and administered by Europol. And it did so for a target that has subsequently been hacked by people who might be ordinary criminals or might be North Korea, using access points (though not the sophisticated techniques) that NSA demonstrated the efficacy of targeting years earlier and which had already been exposed in 2013. Much of the reporting on the SWIFT hack has claimed — based on no apparent evidence and without mentioning the existing, legal TFTP framework — that these hacks were about tracking terrorism finance. But thus far, there’s no reason to believe that’s all that the NSA was doing, particularly with targets like the Kuwait development fund.

Yesterday, the spook site Cipher Brief considered the issue (though mostly by calling on CIA officers rather than NSA hackers).

But I was surprised by a number of things these men (seemingly, Cipher Brief couldn’t find women to weigh in) missed.

First (perhaps predictably given the CIA focus), there’s a bias here on anonymity tied to location, the concern that a hacker might have to be withdrawn, as in this comment from Former Acting Assistant Secretary of Defense for Homeland Defense and Americas’ Security Affairs Todd Rosenblum.

It can lead to the recall of exposed and vulnerable officers that are hard to train and embed in the first place.

And this, from John Sipher.

They can arrest or intimidate the officer, they can kick the officer out of the country or can look to publicly shame or embarrass the officer and his/her country.

But the former NSA spooks who’ve been most vocal about being outed — notably Jake Williams, whom Shadow Brokers exposed even before he released documents with more NSA hackers identified in the metadata, but also Dave Aitel — are concerned about traveling. They largely hacked from the comfort of the US, so being doxed primarily will implicate their freedom of movement going forward (which is directly analogous to Russian hackers, who keep getting arrested while on vacation in US friendly countries). In addition to making vacation planning more complicated, doxing former NSA hackers may limit their consulting options going forward.

These spooks struggle with reciprocity. Consider these two passages in the post:

Russian, Chinese and Iranian governments might seek to retaliate in-kind – which among authoritarian governments often rhymes, rather than duplicates, Western actions.

[snip]

Perhaps most importantly, the intention is part of a larger attempt to create a false moral equivalence between U.S. offensive cyber operations and those perpetrated by adversarial nation-states such as Russia, whose cyber operations leading up Western elections have grabbed the media spotlight.

And this comment from former Chief of Station in Russia Steven Hall:

The Russians live and die by reciprocity. For them, that is one of the linchpins of how they deal with issues like these, and basic diplomatic and policy issues. Typically it has been that if we expel five of their guys, they are going to turn around and expel five of ours. They are always going to look for a reciprocal way to push back. But there are times were they do things that aren’t always clear to us why they consider it reciprocal. And this might be one of those things.

It’s clear they’d like to distinguish what Russia does from what US hackers do. But aside from noting that US doxing of foreign nation-state hackers comes in indictments rather than leaked documents, nothing in this post presents any explanation, at all, about what would distinguish our hackers. That’s remarkable especially since there is one distinction: except where the FBI flips criminal hackers (as in the case of Sabu), our former spook hackers generally don’t use their skills for their own profit while also working for the state. Though perhaps that’s because defense contractors make such a killing in this country: why steal when Congress will just hand over the money?

Other than that, though, I can think of no distinction. And until our spooks and policy makers understand that, we’re going to be the ones impeding any norm-setting about this, not other countries.

But I’m most struck by the rather thin conclusions about the purpose of Shadow Brokers’ doxing, which the post sees as about fear.

If the Shadow Brokers are in fact linked to the Kremlin, then the doxing of NSA hackers is designed to similarly impede current and former U.S. cyber operators from traveling and engaging in clandestine operations abroad – particularly should targeted countries, including allies, take legal action against the individuals for their past involvement in NSA operations. It is also designed to instill fear, as the information could potentially inspire violence against the individuals and their families.

I’m sure the doxing is about fear — and also making it even more difficult for the Intelligence Community to recruit skilled hackers.

But there are at least two other purposes the Shadow Brokers doxing appears to have served.

First, as I noted, the release itself revealed that the US continued to hack SWIFT even after Edward Snowden’s leaks. It hacked SWIFT in spite of the fact that the US has front-door access to SWIFT data under the TFTP agreement with the US. Hypothetically, the US is only supposed to access the data for counterterrorism purposes, but I’ve been assured that the US is in violation of the agreement with the EU on that front. That is, NSA was hacking SWIFT even after the international community had capitulated to the US on access.

By IDing the hackers behind one of the SWIFT hacks, the NSA may have made it easier for other entities to target SWIFT themselves, which has increasingly happened.

More important, still, by doxing NSA hackers, Shadow Brokers likely influenced the direction of the investigation, leading the NSA and FBI to focus on individuals doxed, distracting from other possible modes of compromise (such as the Kaspersky aided third person hacks that appears to have happened with Nghia Hoang Pho and possible even Hal Martin).

More than seven months have passed since Shadow Brokers doxed some NSA hackers, even as he bragged that he had gone nine months by that point without being caught. We still have no public explanation (aside from the Pho plea, if that is one) for how Shadow Brokers stole the NSA’s crown jewels, much less who he is. I’d suggest it might be worth considering whether Shadow Brokers’ doxing — on top of whatever else it did to support Russia’s bid for reciprocity — may have served as incredibly effective misdirection that fed on America’s obsession about insider threats.

On the Timing of the Nghia Hoang Pho Plea

Last Friday, the guy responsible for getting a bunch of NSA hacking tools stolen from his home computer, 67-year old Nghia Hoang Pho, pled guilty to willful retention of classified information. His plea hearing was held in secret; according to the NYT which broke the story, “one courtroom official described the charges against Mr. Pho as ‘super-sealed’ before the hearing.”

According to the information supporting his guilty plea, Pho had been bringing NSA files home for 5 years, from 2010 to 2015.

I want to note something about the timing of the plea. The actual plea deal is dated October 11. It states that “if this offer has not been accepted by October 25, 2017, it will be deemed withdrawn.” The information itself was actually signed on November 29. Friday, the actual plea, was December 1.

So while there’s not a substantial cooperation component in the plea deal, certainly a substantial amount of time took place in that window, enough time to cooperate.

And consider the news coverage that has happened during that period. The initial plea offer was made in the week following a big media blitz of stories blaming Pho (and through him Kaspersky) for the Russian theft of NSA tools. In the interim period between the offer and the acceptance of the plea deal, Kaspersky confirmed both verbally and then in a full incident report that his AV had found the files in question, while noting that a third party hacker had compromised Pho’s machine during the period he had TAO’s tools on it.

In other words, after at least an 18 month investigation, Pho finally signed a plea agreement as the media started blaming him for the compromise of these tools.

During much of that period, Harold Martin was in custody and under investigation for a similar crime: bringing a bunch of TAO tools home and putting them on his computer. Only, unlike Pho, Martin got slammed with a 20-count indictment, laying a range of files, and not just files from NSA. Indeed, the Pho plea notes,

This Office and the Defendant agree that the Defendant’s conduct could have been charged as multiple counts. This Office and the Defendant further agree that had the Defendant been convicted of additional counts, … those counts would not group with the count of conviction, and the final offense level would have increased by 5 levels.

That is, the government implicity threatened Pho to treat him as Martin had been, with a separate charge tied to the individual files he took.

Since April, Martin’s docket has featured continuation after continuation that might reflect cooperation with the government.

All this leads me to believe that these two investigations may have worked in tandem. Whereas the government originally insinuated Martin had provided the files that Shadow Brokers started leaking in August 2016, the Martin cooperation may have led the government to understand the Pho compromise differently. That is, it’s possible that Pho was the source for Shadow Brokers’ tools (or rather, that both men were), but the government didn’t come to understand that until Martin started cooperating.

It’s not clear whether, between the two of them, it would account for all the files that Shadow Brokers had (nor is it clear that Shadow Brokers ever had all the files made available by one or the other of them by loading them onto their home machine). For example, it’s not clear either would have had the San Antonio files at the center of the Second Source theory.

Whatever the details, the timing of the Nghia Hoang Pho plea may suggest that the government only belatedly came to understand how, by loading a bunch of TAO tools running on his Kaspersky-running computer, made the tools available to a third party hack. Certainly, that would explain why Kaspersky has a better understanding of the timing of all this than the government does.