Cyberwar for Sale - Africa Horn Now

MATTATHIAS SCHWARTZ | JAN. 4, 2017 | The New York Times Magazine

After a maker of surveillance software was hacked, its leaked documents shed light on a shadowy global industry that has turned email theft into a terrifying — and lucrative — political weapon.

On the morning of May 18, 2014, Violeta Lagunes was perplexed by a series of strange messages that appeared in her Gmail inbox. It was Election Day to choose the leadership of Mexico’s right-wing Partido Acción Nacional, or PAN, and Lagunes, a former federal congresswoman, was holding a strategy meeting in her office in Puebla city. The emails seemed harmless, at least at first. One appeared to come from the account of a trusted colleague. It asked her to download and review a document. Lagunes clicked on the link, but it seemed to be broken, so she wrote back to her colleague and asked him to send it again. Elsewhere in her inbox was an email from Google warning her that someone had tried to log in to her account. Meanwhile, she began to receive phone calls from PAN allies, who claimed that they had received emails from Lagunes’s account that she did not remember sending.

Now Lagunes was worried. Around 1 o’clock, she called the colleague who appeared to have emailed her. She reached him at a restaurant, where he was finishing lunch with other campaign allies. “I did not send you an email,” he insisted. A consultant with the campaign — who asked to remain anonymous in order to preserve his relationships with other candidates — overheard the conversation. He knew of other campaign workers who had been receiving similar messages: emails with vague subject lines, asking the recipient to review a document or click a link. The campaign, he realized, had been hacked.

In the vote for party leader, Lagunes and her allies in Puebla — a two-hour drive southeast from Mexico City — were supporting the challenger, a senator who promised to return the party to its conservative roots. But the incumbent was backed by Puebla’s powerful governor, Rafael Moreno Valle. One of Mexico’s rising political stars, Moreno Valle is close to Mexico’s president, Enrique Peña Nieto, and has forged an alliance between PAN and Nieto’s centrist Partido Revolucionario Institucional, or PRI, long the dominant force in Mexican politics. Since winning the governorship in 2010, Moreno Valle’s opponents say, his ambitions have grown, and he has resorted to increasingly harsh measures to keep Puebla state — including members of his own party — under control. “In the beginning, the governor was low-profile and respectful,” Rafael Micalco, a former leader of PAN in Puebla state, told me. “When he became governor, he transformed. Now he controls the party through threats.”

This race to retain control of the party leadership in 2014 was a crucial test for the governor, who was rumored to be considering a run for Mexico’s presidency in 2018. (This past September, Moreno Valle publicly announced his intent to run.) Clashes between the two camps were especially intense in Puebla, where backers of the challenger, Ernesto Cordero, claimed that the governor was using public money to support the incumbent, Gustavo Madero, though the governor’s office has denied these charges. Shortly before the election, Madero’s campaign manager said that Cordero’s side was trying to undermine the legitimacy of the process. “Their strategy is clear from the outset,” he said in an interview with a Mexican magazine. “ ‘If I win, good. If not, I was cheated.’ ”

“The day before,” the consultant told me, the field network was “motivated and eager to do this work. After the hack, it was very hard to reach them. The few who did answer said that they had received phone calls saying that their lives were at stake. They were worried that if they went out, they or their families would get hurt.”

According to another worker on Cordero’s campaign, who also requested anonymity, citing fear of reprisal, the message to the canvassers was simple and direct: “We know who you are. If you don’t want any trouble, shut down your cellphone and stop your activity.” The worker added: “It’s an authoritarian regime.”

Madero won the election, with 57 percent of the 162,792 votes cast over all. In Puebla, his margin was substantially larger, roughly 74 percent. Cordero’s team decided not to contest the result. They had suspicions about how they were hacked. But it would be another year before any evidence emerged. Their political enemies, leaked documents seemed to show, had built a spying operation using software made by an Italian firm called Hacking Team — just one of many private companies that, largely below public notice, have sprung up to aid governments in surveilling the private lives of individual citizens. The industry claims that its products comply with local laws and are used to fight crime and terror. But in many countries around the world, these tools have proved to be equally adept at political espionage.

On average, an American office worker sends and receives roughly 120 emails per day, a number that grows with each passing year. The ubiquity and utility of email has turned it into a fine-grained record of our day-to-day lives, rich with mundane and potentially embarrassing details, stored in a perpetual archive, accessible from anywhere on earth and protected, in some cases, by nothing more than a single password. In the case of Violeta Lagunes, her email login represented a point of vulnerability, a seam where the digital walls protecting her campaign were at the mercy of her human judgment — specifically, whether she could determine if a message from an apparently reputable source was real or fake. Nearly two years later, John Podesta, chairman of Hillary Clinton’s campaign, was faced with a similar judgment call. An email warned him that someone in Ukraine had tried to access his Gmail account and asked him to click on a button and reset his password. His senior adviser forwarded the email to one of the campaign’s technology experts. “This is a legitimate email,” he replied, in what the expert later would clarify was a simple typing error on his part; he meant to say it was not legitimate. “The gmail one is REAL,” the senior adviser wrote to Podesta and another aide.

And so, like Lagunes, Podesta fell into a trap. The button appeared to lead to an official Google page, but it was in fact a meticulously personalized fake, with a domain address linked to a remote cluster of atolls in the South Pacific. The details were designed to trick Podesta into entering his password. This technique is known as “spear phishing.” It is an especially potent weapon against companies and political organizations because it needs to succeed only one time, against one target. After that, attackers can use the trusted identity of the first compromised account to more easily lure colleagues into opening infected attachments or clicking on malicious links. Not only will a working email password yield years of intraoffice chatter, invoices, credit-card bills and confidential memos; it can often be leveraged into control of other personal accounts — Twitter, Facebook, Amazon — and even access to company servers and internet domains.

The Podesta and Lagunes episodes are far from the only cases in which hackers have used information from stolen emails as a weapon against an entire institution. The 2009 “Climategate” incident, which exposed troves of emails from prominent climate researchers, began when hackers remotely broke into servers at a British university with the help of illicitly obtained passwords. The 2014 hack of internal Sony files, which American officials attributed to the North Korean government, began with a series of spear-phishing emails that attackers then used to dig deeper into Sony’s servers. Each hack yielded the most private thoughts and deeds from the members of each respective organization: their blunt insults, their quashed dissents, their half-baked plans, their smarmy flattery, all chronicled in time down to the hundredth of the second when the author clicked “send.” In an earlier era, the hackers might have had to engage in riskier behavior, like bribery or burglary. Now, in many cases, all they had to do was send along a link.

The White House, C.I.A. and F.B.I. have all claimed that, based on classified evidence, they can trace the hacks of Podesta’s email account (and other hacks of people close to the Clinton campaign) back to the Russian government. But with the rise of private firms like Hacking Team, penetrating the email accounts of political opponents does not require the kind of money and expertise available to major powers. A subscription-based website called Insider Surveillance lists more than a dozen companies selling so-called ethical malware, including Milan-based Hacking Team, the German firms FinFisher and Trovicor and the Israeli company Nice. Compared with conventional arms, surveillance software is subject to few trade controls; a recent attempt by the United States to regulate it under a 41-country pact called the Wassenaar Arrangement failed. “The technology is morally neutral,” says Joel Brenner, a former inspector general of the National Security Agency. “The same program that you use to monitor your babysitter might be used by Bashar Assad or Abdel Fattah el-Sisi to keep track of whomever they don’t like.”

Hacking Team has fewer than 50 employees, but it has customers all over the world. According to internal documents, its espionage tool, which is called the Remote Control System, or R.C.S., can be licensed for as little as $200,000 a year — well within the budget of a provincial strongman. After it has been surreptitiously installed on a target’s computer or phone, the Remote Control System can invisibly eavesdrop on everything: text messages, emails, phone and Skype calls, location data and so on. Whereas the N.S.A.’s best-known programs grab data in transit from switching rooms and undersea cables, the R.C.S. acquires it at the source, right off a target’s device, before it can be encrypted. It carries out an invisible, digitized equivalent of a Watergate-style break-in.

The United States government is almost certainly the world’s most formidable repository of hacking talent, but its most powerful cyberweapons are generally reserved for intelligence agencies and the military. This might explain why, according to company documents, at least two federal law-enforcement agencies have been Hacking Team clients: the F.B.I., beginning in 2011, and the Drug Enforcement Administration, beginning in 2012. The F.B.I. contract paid Hacking Team more than $700,000; the D.E.A. appears to have used the software to go after targets in Colombia.

Documents show that the company has also sold its software to some of the world’s most repressive governments. Some, like those of Honduras, Ethiopia, Bahrain, Morocco, Egypt and Saudi Arabia, are Western allies. Other countries, like Uzbekistan and Turkey, have a more troubled relationship. A few are openly hostile to the West. Between 2012 and 2014, Hacking Team was paid nearly one million euros by the government of Sudan, a United States-designated state sponsor of terrorism. Even more notable, in light of recent events, is the three-year relationship that Hacking Team carried on with the F.S.B., one of Russia’s main intelligence agencies. As with Puebla, Hacking Team used a middleman, a research agency called Kvant, to handle its sales to Russia. Between 2012 and 2014, the agency paid Hacking Team 451,000 euros to license the Remote Control System.

Violeta Lagunes, whose email account was compromised during a Partido Acción Nacional election campaign. Credit Dominic Bracco II/Prime, for The New York Times

After Lagunes’s call on Election Day, her colleagues rushed from the restaurant back to their local headquarters, a hotel conference room that they had nicknamed “the bunker.” All morning, they had been trying to reach their field network, a group of 40 Cordero canvassers who were working to get out the vote in Puebla state. But the field network seemed to have gone dark. Few of the canvassers were even answering their phones. Hackers, the team concluded, must have found the list of the canvassers’ names and phone numbers — widely circulated by email within the campaign — and begun to intimidate them.

Hacking Team claims that it draws the line at customers who commit “gross human-rights abuses” and that it sells exclusively to governments operating within the laws of their own countries. In at least one case, David Vincenzetti, Hacking Team’s founder and chief executive, told a salesman to hold off on a potential Mexican client. “We sell to official, governmental LEAs” — law-enforcement agencies — “and security agencies ONLY,” Vincenzetti wrote in an email. But at other times, a more casual attitude prevailed inside the company. “If one sells sandwiches to Sudan, he is not subject, as far as my knowledge goes, to the law,” one Hacking Team lawyer wrote in an internal email. “Hacking Team should be treated like a sandwich vendor.”

When asked about its arrangements in various countries, the company responded that it “does not comment on confidential business dealings.” Its American spokesman, Eric Rabe, did tell me that neither Russia nor Sudan is a current Hacking Team customer. (The relationships, Rabe wrote, ended in 2014, Russia because “the Putin government evolved from one considered friendly to the West to a more hostile regime” and Sudan “because of concerns about the country’s ability to use the system in accordance with the H.T. contract.”) Separately, the company confirmed that the state of Puebla was, in fact, a former client.

Until recently, most of what was known about the world of private surveillance companies was a matter of hearsay and speculation. Industry players kept a low profile, operating discreetly from rented offices and meeting potential customers in person a few times a year at carefully screened trade shows. This is why it was so notable when, in July 2015, an unusual tweet appeared in Hacking Team’s Twitter feed. “Since we have nothing to hide,” it read, “we’re publishing all our emails, files and source code.” Then came another tweet, with links to a downloadable file called Hacked Team. The file was huge, 420 gigabytes of material scraped from Hacking Team’s internal servers. Inside were 33 folders containing the company’s contracts, payroll documents, invoices, legal memos, customer-support records and a five-year cache of email correspondence from the chief executive on down.

Hacking Team had itself been hacked. WikiLeaks pounced on the breach and quickly uploaded the emails into a searchable database. Anyone with an internet connection could now read the chief executive joking about how his company was in the business of selling “the evilest technology on earth.” You could browse Hacking Team’s source code, including one line using “bomb_blueprints.pdf” as a placeholder for files that might be found on a target’s device. On Reddit, an online peanut gallery formed around the time-wasting online habits of one Hacking Team engineer and his own notably weak passwords — HTPassword!, P4ssword, Passw0rd.

But the most damaging exposures in the leak by far were Hacking Team’s client list and the names of some of the clients’ targets. In South Korea, newspapers focused on evidence suggesting that Hacking Team’s software had helped the nation’s intelligence service rig an election; after the leak, one agent who had reportedly used the system there committed suicide. In Ecuador, a magazine found an email with seven phone numbers that the government appeared to have targeted with the R.C.S. Three belonged to lawmakers; a fourth to the mayor of Quito; all four were members of the opposition party.

With the source code for the Remote Control System now public, the company and its clients had to stop using it temporarily. By the end of the year, though, Hacking Team had updated its product and was trying to rebuild its reputation. I was curious whether a company that profited from online breaches could recover from its own. Eric Rabe, the American spokesman, sounded eager to meet me for coffee in Philadelphia. A grandfather and former television-news anchor, he exudes square-jawed credibility. “If you disagree with someone on the internet,” he said, of the Hacked Team files, with a wave of his hand, “there’s no need to have a public-policy debate. Just go destroy them.”

A couple of months later, I went to Milan to visit Hacking Team’s headquarters, a stately gray apartment building with boxes of limp flowers adorning a few of its sooty sills. Waiting to demonstrate the company’s software were Rabe; Philippe Vinci, a company vice president; and Alessandro Scarafile, a young engineer. Scarafile had gathered a Dell desktop computer and three smartphones: iPhone, BlackBerry and Android. The screen from his own laptop, which represented the console of a client intelligence agency, was projected on the wall. Several icons represented the various streams of data that could potentially be acquired by gaining control of the target’s computer: images from built-in cameras, sound from built-in mikes, screenshots, detailed records of applications opened and bitcoins transferred, a continuous log of location with latitude and longitude, and logs of address books, calendars, phone calls, Skype calls and passwords, as well as websites visited. A key logger recorded every key that was pressed. It was a lot to keep track of. Two other views, called “line of events” and “line of actions,” assembled the information into chronological order.

Scarafile, who was playing both the customer and the target, or “bad guy,” turned on the Dell. Judging by the background on the desktop, a gothic scene framed by castlelike silhouettes, our target seemed to be planning a terrorist attack from Transylvania.

There are three methods, Scarafile explained, for getting the Remote Control System onto a target’s device. Customers can gain physical access to the device and then infect it with a USB stick or memory card. They can beam the R.C.S. in over a Wi-Fi network. Or they can send the customer an email and get him to click on an infected attachment — usually a file from a brand-name program like Microsoft Word or PowerPoint. Scarafile did not mention a fourth method, one described by Hacking Team’s critics and referred to in its internal emails: the installation of the R.C.S. through a more elaborate process called “network injection,” which is said to involve pinpointing the target’s exact location on the internet, observing the person as he orders up, say, a YouTube cat video, and then serving up a doctored version of the same page, one with the desired cat video playing in the foreground — YouTube logo and all — as the R.C.S. discreetly rolls in past the digital gates.

Whatever the method of infection, the malicious code, known as an “agent,” then communicates with its masters anonymously, its dispatches routed through a series of dedicated servers scattered around the world. Even if the target gets suspicious and figures out that something is wrong, this chain of servers makes it nearly impossible for him to figure out exactly who is using Hacking Team’s product to spy on him.

David Vincenzetti, Hacking Team’s chief executive, in Milan. Credit Luca Locatelli/Institute, for The New York Times

For the demo, Scarafile opted for the third method, which he referred to as “a bit of social engineering.” Using the customer’s computer, he sent the target an email with a Word file attached. Then he returned to the target’s computer and double-clicked on it, just as Lagunes had clicked on the link in her inbox. “From now on,” he said, “this system is currently infected, or monitored, by the Remote Control System.” It would remain so even if the suspect turned off his machine or logged out.

On the timeline, the target appeared as a stubbly lout in an undershirt, named “Jimmy Page … head of the terrorist cell.” Using Jimmy Page’s Dell, Scarafile logged into Page’s Gmail, Facebook and Twitter. He opened up Skype, perused Page’s criminal colleagues (Don Corleone, Harry Potter, Keyser Soze) and left one of them a voice mail message from Page’s phone. Accessing a USB drive attached to the infected computer, Scarafile opened an encrypted file that turned out to contain an order to “kill David Vincenzetti.” R.C.S. captured it all, including periodic snapshots of Scarafile, as Page, at work.

“I don’t like the words ‘inject’ and ‘infect,’ ” Vinci, the vice president, said. “R.C.S. is deploying the agent into the device of the target because you want to monitor some of his activities. Exactly the way that law enforcement is listening to some of your phone calls, right?”

Rabe had told me that Vincenzetti was “a fighter,” and his combative side was apparent from a collage of magazine clippings and printouts taped to the wall beside his desk — a sort of mood board, like those used in the advertising industry to gather inspiration before the unveiling of a new brand. Vincenzetti’s mood board, though, wasn’t about sneakers or cola. It was about the global struggle for power, which he seemed to envision as one big conflict, a battle between the good guys and the bad guys. On the good side was a photo of the bell that aspiring Navy SEALs can ring should they want to quit the program during Hell Week, and a quote from another entrepreneur that “business is war.” On the bad side, Vincenzetti had taped up a satellite image showing one of the hottest zones of international tension, the artificial islands rising in the South China Sea, a line of terrestrial pawns advancing China’s sphere of influence. Beside it was a chart about Iran, depicting how the country could continue to advance its nuclear program despite the recent American-led deal.

“I think the Iran deal is just terrible,” Vincenzetti said, and then added, sarcastically, “Oh, it’s such a very peaceful expansion. Very peaceful.”

“That is not an official company position,” Rabe interjected.

Vincenzetti, now 48, is a familiar type — a ferociously competitive, driven entrepreneur whose existence is organized around his work. He has a wife, who was born in Morocco, and no children. His small, heavy-lidded eyes can make him seem sleepily blasé. On the morning we met at Hacking Team’s offices, he wore jeans, a navy-blue cardigan and a striped shirt, unbuttoned to midchest. He seemed more comfortable on his feet than sitting down. As we talked in a conference room, he periodically leapt to his feet and stalked around the table, considering in turn the espresso machine, the view from the window, a case of bottled water.

“If I wanted to break into this room, how would I do it?” he asked. “There is a door, and there are two windows.” He pressed his hands against the glass panes. “The perimeter is the first thing you must secure,” he continued. Securing data was what he did earlier in his career. Now he had moved on. “If you cannot break into a bank, you cannot protect a bank. So when you are in security, really there is no difference between thinking offensively and defensively.”

In the mid-1980s, Vincenzetti’s parents, a salesman and a schoolteacher, bought him a Commodore 64, one of the earliest personal computers. He soon created a Pac-Man clone, a Tron lightcycle-style game and a text-based adventure game. As a computer-science student at the University of Milan in the ’90s, he became fascinated by cryptography; he corresponded with programmers around the world about new cryptographic theories and wrote code for email encryption. In his senior year, he was appointed to administer the university’s internal network, a post that was usually reserved for a graduate student. Vincenzetti remembers these early days as a time when “everything was free and no one was trying to harm you. We were wide open and accessible. All the best were called hackers, and I was a hacker.”

Vincenzetti left university early and founded three companies, all of them focused on defensive cybersecurity. After he founded Hacking Team in 2003, he tried to sell his services to Italian police agencies but found them skeptical that Mafiosi and other high-level criminals would ever bother to encrypt their communications. In Italy, the police were also used to getting whatever they needed through wiretaps, arranged with varying degrees of formality through their contacts at telecommunications firms. But after the 2004 Madrid train bombings, which were coordinated via cellphones and the internet, police officers and intelligence agents not just in Italy but all across Europe became interested in contracting with offensive-hacking vendors, part of an emerging arms race over consumer-grade encryption. The growth of Skype made it easy for users to encrypt their communications, and the authorities were eager to pay for countermeasures like the Remote Control System. Singapore, Hacking Team’s first non-European client, signed on in 2008. The company’s Middle Eastern business took off in 2011, a boom that coincided with the beginning of the Arab Spring.

An entrance to Hacking Team’s Milan offices. Credit Luca Locatelli/Institute, for The New York Times

By then, Hacking Team had entered a growth phase, its business driven in part by demand among third-world governments for first-world surveillance tools. According to two former employees, the company held talks with Col. Muammar el-Qaddafi’s chief security officer, who wanted to build a countrywide bulk-collection scheme that could be embedded in every Libyan cellphone. (Rabe would not confirm or deny that this meeting occurred, and added: “The company often receives requests to provide services that it does not have available or would sell.”) The United Nations, which prohibits the export of “electronic weaponry” to Sudan, has investigated Hacking Team’s activities there. Nineteen members of the Italian Parliament signed a petition raising the question of whether the Egyptian government might have used the R.C.S. to track Giulio Regeni, a 28-year-old Italian student who appears to have been under government surveillance and whose mutilated body turned up on the side of a road in Egypt last year. Hacking Team’s software has not been connected to the case, but the company has done business with the current Egyptian regime. As part of the controversy, the Italian government temporarily revoked Hacking Team’s global export license, so that for several months the company had to file a separate application for each of its customers outside the European Union. (When I brought up the Regeni case with Rabe, he called the dead student “this Italian national who got himself killed in Egypt.” He cited Hacking Team’s official policy — the company neither knows nor desires to know the identities of the people its customers choose to target. “There is no evidence that Hacking Team’s software had anything to do with Regeni’s demise,” Rabe said.)

All Hacking Team customers sign contracts agreeing to comply with local laws. The company says that it vets potential customers and studies reports from journalists and human-rights groups, looking for “objective evidence or credible concerns” that its products are being abused. But when it comes to Hacking Team’s own interactions with customers, leaked documents suggest that employees have sometimes turned a blind eye. In the case of the Puebla government and other Latin American customers, Hacking Team employees appeared to ignore warnings suggesting that the Remote Control System was being used to gather intelligence on the political opposition. On multiple occasions, customers emailed Hacking Team attachments with election-related content, including polling data, party registration forms and invitations addressed to and signed by elected officials. Rather than ask what these files had to do with fighting crime and drug trafficking, Hacking Team support-staff members simply emailed them back, as requested, with an embedded “exploit,” turning the document into a surveillance tool to be used against whomever it was sent on to. Asked about these cases, Rabe replied that customers “are not supposed to be using it for political purposes, but I don’t think it’s reasonable to expect that an Italian computer programmer,” i.e., a support technician, “would have seen these files and known what was going on. … I think that’s a stretch, that an Italian software guy could know that an individual is a dissident.”

Hacking Team’s most persistent critic is Citizen Lab, a research group at the University of Toronto’s Munk School of Global Affairs. Before the Hacked Team leak, Citizen Lab documented cases in which Hacking Team software turned up on the devices of activists in Morocco and the United Arab Emirates, as well as an Ethiopian-American journalist in Alexandria, Va. Ronald Deibert, Citizen Lab’s director, told me that Hacking Team “is a company that appears to have no internal controls on abuse of its products.” When I asked Vincenzetti about this, he said that Citizen Lab was motivated by money, noting that the group won a million-dollar grant a week after publishing a report on Hacking Team’s sales to Ethiopia. “Their identity,” he said, “is: ‘I am the defender of free speech, I am the defender of liberty and democracy.’ O.K. So am I. So is every rational guy.” If Citizen Lab really cared about good and evil, he said, it would be fighting China and Iran.

Almost immediately after the Hacked Team documents went online, they were being pored over by R3D, a Mexico City-based civil-liberties group. Luis Fernando García, R3D’s director, says that intimidation and online surveillance have increased under Peña Nieto’s presidency, and he took notice when Citizen Lab, in 2014, released a report that traced a chain of servers associated with Hacking Team that routed data through Hong Kong, London, Amsterdam and Atlanta before terminating somewhere inside Mexico. Exactly who was on the receiving end of the traffic and what they were doing with it was unknown at the time, but now R3D’s team recognized that this could be their chance to find out.

Shortly thereafter, R3D published three invoices from the Hacked Team cache showing that one state government — that of Jalisco, on the Pacific Coast — had paid the company nearly half a million euros for the Remote Control System. Soon other documents were found in the leak to implicate several more states, including Puebla, though most of the states denied ever using the software. (Jalisco later admitted to purchasing the system, which it claimed was for its prosecutor’s office.) The story failed to get much traction in the Mexican press, except in Puebla, after R3D connected with Lado B, a small online-news collective there. The site’s name translates to “B Side,” as in the back of a hit single — symbolizing its dedication to telling stories that would otherwise go untold.

Lado B’s editor, Ernesto Aroche, was not surprised that Puebla’s government had been using the Remote Control System. In Moreno Valle’s six years as governor there, he has spent lavishly on new surveillance systems, including multiple “security arches,” highway-spanning structures that scrutinize traffic with video cameras and X-rays, and whose cost had raised some questions in the local press. Beginning in 2013, Aroche began noticing solicitations on the Puebla-state website for hidden cameras and other spying equipment. When he filed a public-records request about them, the government’s form response indicated that these orders had never been filled, but Aroche was skeptical, given the mounting evidence that Moreno Valle’s administration had begun using its security apparatus for political purposes. There were periodic break-ins of homes belonging to dissident politicians and journalists, with the burglars sometimes taking little besides their victims’ laptops. Politicians would answer calls from unknown numbers, only to hear their own taped conversations played back to them. Another journalist, Fernando Maldonado, received an unmarked envelope full of what were purportedly transcripts of 400 private phone calls made by Puebla politicians.

As Aroche and R3D sifted through the Hacked Team file dump, they discovered that the files on Puebla had more detail than those dealing with Hacking Team’s other Mexican accounts. The Puebla client often wrote emails asking for help with infecting a particular document with a malicious virus. Some of these came from an account — soporteuiamx@gmail.com — that also appears in many of Hacking Team’s internal support tickets. One of those tickets documents a client’s request for help in infecting an attachment: an invitation to attend an event for a Mexican political organization, signed by Violeta Lagunes and addressed to another party colleague opposed to the governor.

Aroche interviewed politicians and journalists who were on the receiving end of the infected emails contained in the Hacked Team trove. According to a 2015 article that Lado B published in conjunction with a politics website, the Puebla group sent Hacking Team at least 47 requests to infect specific files that it would then forward on to their targets. Almost all of those files had to do with political issues. Going through the Hacked Team materials, Aroche found the name of a Puebla government employee who seemed to be working within the spying operation. “Before, we had been talking about ghosts,” Aroche told me. “Now, we could prove it. We started putting names on the actors.”

More than a year later, there have been no political repercussions for the governor or his allies in Puebla. I met Aroche this fall at the Lado B offices, a small room facing a courtyard in a crumbling stone building from Puebla’s colonial days. The only suggestion of Lado B’s presence was a small sticker on the window. Inside, Aroche was thumbing through an envelope of documents that he had wrung out of the state government. Two redacted contracts signed by Puebla officials showed an arrangement between the state government and a company called Sym Servicios Integrales, which Aroche’s reporting had identified as an intermediary for Hacking Team. The details of what had been purchased were redacted. “These contracts demonstrate that the Puebla government had a commercial relationship with this company,” Aroche said. “I’m sorry if they’re a little bit stained — I dropped some coffee on them.” (Sym Servicios Integrales says it “never sold H.T. technology to the state of Puebla.”)

In a brief written statement from Sagrario Conde Valerio, a spokeswoman, the government of Puebla denied allegations of spying and claimed that “no relationship exists nor has ever existed between the government of Puebla and the company ‘Hacking Team.’ ” The government declined to respond further to a list of questions.

But a former official from Moreno Valle’s administration claimed to me that more than $1 million was being diverted from the state budget each year to fund the political-espionage unit. Documents from the Hacked Team archive indicate that Hacking Team received an order from the government of Puebla for 415,000 euros in the spring of 2013, and that Hacking Team booked hotel rooms for three of its associates who traveled to Puebla in May of that year.

‘Before, we had been talking about ghosts. Now, we could prove it. We started putting names on the actors.’

Some of the R.C.S. training, I was told, took place in a green building on a residential street. When I visited the purported site this fall, lettering above the door said that the building was once a school. Now it had bars across the door and mirrored windows. Neighbors told me it had been abandoned for several months. “People would come and go,” one woman said. “They were very secretive. They would drop off equipment, take away equipment. Then one day, about a year ago, they came and took everything away and left.” Another neighbor said that he often saw a state-police car parked in front of the house and a man with a limp entering and leaving — a former Mexican intelligence agent, according to someone who was present during the Hacking Team training.

Mexico is Hacking Team’s biggest export market, accounting for nearly six million euros in sales, according to leaked documents. Ostensibly, the Remote Control System is intended for fighting criminals and drug traffickers there. (“There have been reports that the software was used in the apprehension of Chapo Guzman,” Rabe told me, referring to the Mexican drug lord. “I can’t confirm it.”) The files indicate that at least seven other Mexican state governments were Hacking Team clients, but because they did not use email to the same extent as Puebla, their activities are harder to track. Multiple former Hacking Team employees told me that abuses of the software were not limited to Puebla. One former employee described having the system set up inside a mayor’s office.

Katitza Rodríguez, international-rights director at the Electronic Frontier Foundation, says that Mexican law allows for interception of communication, like the wiretapping of phone lines, but it does not grant precise legal authority to use new, powerful invasive tools like those created by Hacking Team. And she argues that it is much more dangerous in Mexico than in other Western countries, where checks and balances are stronger. “This is much more intrusive than the interception of a phone call,” she said. “They are not only listening; they are taking over your laptop. Mexico needs to have a full debate in Congress about what legal safeguards are needed for this kind of surveillance or if the government should be using it at all.”

The Hacked Team files indicate that in mid-2015, Hacking Team opened an American subsidiary and leased office space in Reston, Va., a 20-minute drive from the C.I.A.’s headquarters. The company drew up a “U.S. action plan,” calling for a significant North American expansion, with new hires and rounds of pitches to the Department of Justice, the United States military and the Royal Canadian Mounted Police. Hacking Team was also targeting another potentially lucrative market: American state and local governments. It pitched the R.C.S. to law-enforcement agencies in San Bernardino, Calif.; Washington; New York; Fort Lauderdale; and Orlando. In a risk assessment commissioned by the company, lawyers advised that such sales were probably legal, so long as the product was provided “at a distance” from actual investigations and used “in the normal course of government operations.”

American expansion by companies like Hacking Team may not face much resistance from the federal government, which is becoming more accepting of electronic surveillance as part of normal police work. Last year, the Justice Department successfully changed the rules of criminal procedure, making it easier for federal agents to hack into multiple computers with a single warrant. Senator Steve Daines of Montana, who tried to block the change, complained that it gives “unlimited power for unlimited hacking.” In a 2014 address, now referred to in cybersecurity circles as his “Going Dark” speech, the F.B.I.’s director, James Comey, argued that the encryption built into Apple products posed a threat to public safety — enabling not only terrorism but also drug trafficking, child abuse and hit-and-runs. Instead of “a safe that can’t be cracked,” Comey wanted technology companies to lend him the combination. “The law hasn’t kept pace with technology, and this disconnect has created a significant public-safety problem,” Comey said. “We call it ‘going dark.’ ”

As the United States government has argued for weaker protections around personal communications, American businesses have been spreading more powerful surveillance tools around the world. Many foreign law-enforcement agencies already buy electronic-surveillance products from SS8 — a company backed by the storied venture-capital firm Kleiner Perkins Caufield Byers — as well as from the Harris Corporation, a $13 billion company based in Florida and traded on the New York Stock Exchange. At least 25 American police departments have used Harris’s Stingray device, which mimics cellphone towers and can intercept cellphone calls within 200 meters. All known sales of Stingrays to local law enforcement have been authorized by the F.B.I., which has since fought in court to keep those sales secret. In Baltimore alone, Stingrays have been used more than 4,000 times, in routine drug investigations. Much less is known about another Harris product, the advanced Hailstorm, which is also sold to local police departments. Hailstorm is reportedly capable of implanting malware that can take control of a targeted phone, much like Hacking Team’s R.C.S.

The Hacked Team documents that offer the most revealing view of the company’s ethos happen to be the most public ones. For years, as often as two or three times a day, Vincenzetti sent mass emails to hundreds of his business contacts. The recipients included numerous members of the United States military and intelligence community, as well as government employees from the city of Cincinnati and the Internal Revenue Service. In these messages, Vincenzetti often addresses this audience collectively as “gents.” The news he cites is a reminder of how the geopolitical winds have been blowing in favor of Hacking Team and other self-described allies of law and order. In Vincenzetti’s world, the system is always, as George Tenet famously said about pre-Sept. 11 intelligence, “blinking red”: the imploding Middle East; a restive, nuclear-armed Russia; battalions of ISIS-trained jihadis roaming around Europe with their encrypted thumb drives and dark-web expertise. Against this backdrop of ever-increasing danger, concerns about human rights are naïve at best.

Vincenzetti’s emails vividly exploit this sense of danger and alarm. He writes about shadowy gangs of Iranian hackers using the #JeSuisCharlie hashtag to inject malware into French laptops. He celebrates the conviction of Ross Ulbricht, a.k.a. the Dread Pirate Roberts, creator of the Silk Road website. After the arrests of two Uzbek men in Brooklyn for telling informants that they wanted to join up with ISIS, Vincenzetti writes of “a very serious terrorist plot on American soil foiled.”

He went on to tap out this sales pitch:

The time has come for a technologically MORE SOPHISTICATED, and much more effective, internet supervision … something capable to penetrate the core of the terrorists’ HIDDEN forums. And such a (quite unique) technology EXISTS.

The following day, he gave a few more hints:

I am talking about a NEW technology capable of neutralizing their encryption-based protective layers in order to track them, identify them, locate them, chase them and finally bust them. Something operating on a massive scale. Something different. I am talking about a novel, superior, next-generation mass-surveillance technology.

The tone of Vincenzetti’s sales patter was strangely upbeat, especially considering his dire forecasts. It was almost as if he were in the business of selling microwave ovens or sandwiches, not tools through which the private lives of criminals (and whoever else) could be fully laid bare.

Vincenzetti’s unstated equation — privacy is secrecy, and secrecy is terrorism — is less controversial than it might appear. A supportive echo can be heard in many public statements from American officials, which Vincenzetti often cut and pasted into his mass emails. The former attorney general Eric Holder called for “investigative and prosecutorial tools that allow us to be pre-emptive.” When Comey warned that “encryption threatens to lead all of us to a very dark place,” Vincenzetti forwarded it along approvingly, with the tagline: “We DO have an answer to many if not all of his concerns.” And when, in May 2015, Comey warned of a “threat” that had “morphed” into “a chaotic spider web,” Vincenzetti sent word to his “gents” as well.

One month later, an anonymous hack revealed Hacking Team’s own invisible spider web, and one year later, during the run-up to Election Day, came the internal emails of the Democratic Party. By then it was clear that tools for digital burglary had spread well beyond the hands of regular police officers. Comey had argued for weaker safes; Vincenzetti was selling longer crowbars. They could be used to chase Jimmy Page, arrest Chapo Guzman, fight crime, smear a political opponent or just keep tabs on someone, anyone. To be hidden is to be a terrorist — this was the heart of his pitch. Any digital redoubt that could resist being pried open was a public risk and a private opportunity.