ON March 11, 2013, Thomas Donilon, President Obama’s national security adviser, gave a speech at the Asia Society on Manhattan’s Upper East Side. Much of it was boilerplate: a recitation of the administration’s policy of “rebalancing its global posture” away from the ancient battles of the Middle East and toward the “dynamic” region of Asia-Pacific as a force for growth and prosperity.
But about two thirds of the way through the speech, Donilon broke new diplomatic ground. After listing a couple of “challenges” facing U.S.-China relations, he said, “Another such issue is cyber security,” adding that Chinese aggression in this realm had “moved to the forefront of our agenda.”
American corporations, he went on, were increasingly concerned “about sophisticated, targeted theft of confidential business information and proprietary technologies through cyber intrusions emanating from China on an unprecedented scale.”
Then Donilon raised the stakes higher. “From the president on down,” he said, “this has become a key point of concern and discussion with China at all levels of our governments. And it will continue to be. The United States will do all it must to protect our national networks, critical infrastructure, and our valuable public and private sector property.”
The Obama administration, he said, wanted Beijing to do two things: first, to recognize “the urgency and scope of this problem and the risk it poses—to international trade, to the reputation of Chinese industry, and to our overall relations”; second, to “take serious steps to investigate and put a stop to these activities.”
The first demand was a borderline threat: change your ways or risk a rupture of our relations. The second was an attempt to give Chinese leaders a face-saving way out, an opportunity for them to blame the hacking on hooligans and “take serious steps” to halt it.
In fact, Donilon and every other official with a high-level security clearance knew that the culprit, in these intrusions, was no gang of freelance hackers but rather the Chinese government itself—specifically, the Second Bureau of the Third Department of the People’s Liberation Army’s General Staff, also known as PLA Unit 61398, which was headquartered in a white, twelve-story office building on the outskirts of Shanghai.
Since the start of his presidency, Obama had raised the issue repeatedly but quietly—in part to protect intelligence sources and methods, in part because he wanted to improve relations with China and figured a confrontation over cyber theft would impede those efforts. His diplomats brought it up, as a side issue, at every one of their annual Asian-American “strategic and economic dialogue” sessions, beginning with Obama’s first, in 2009. On none of those occasions did the Chinese delegates bite: to the extent they replied at all, they agreed that the international community must put a stop to this banditry; if an American diplomat brought up China’s own involvement in cyber hacking, they waved off the accusation.
Then, on February 18, Mandiant, a leading computer-security firm, with headquarters in Alexandria, Virginia, published a sixty-page report identifying PLA Unit 61398 as one of the world’s most prodigious cyber hackers. Over the previous seven years, the report stated, the Shanghai hackers had been responsible for at least 141 successful cyber intrusions in twenty major industrial sectors, including defense contractors, waterworks, oil and gas pipelines, and other critical infrastructures. On average, these hackers lingered inside a targeted network for a full year—in one case, for four years and ten months—before they were detected. During one particularly unimpeded operation, they filched 6.5 terabytes of data from a single company in a ten-month period.
Kevin Mandia, the founder and chief executive of Mandiant, had been one of the Air Force cyber crime investigators who, fifteen years earlier, had nailed Moscow as the culprit of Moonlight Maze, the first serious foreign hacking of Defense Department computers. Mandiant’s chief security officer, Richard Bejtlich, had been, around the same time, a computer network defense specialist at the Air Force Information Warfare Center, which installed the first network security monitors to detect and track penetrations of military computers. The monitoring system that Mandia and Bejtlich built at Mandiant was based on the system that the Air Force used in San Antonio.
While putting together his report on Unit 61398, Mandia was contracted by The New York Times to investigate the hacking of its news division. As that probe progressed (it turned out that the hacker was a different Chinese government organization), he and the paper’s publishers discussed a possible long-term business arrangement, so he gave them an advance copy of the report on the Shanghai unit. The Times ran a long front-page story summarizing its contents.
China’s foreign affairs ministry denounced the allegation as “irresponsible,” “unprofessional,” and “not helpful for the resolution of the relevant problem,” adding, in the brisk denial that its officials had always recited in meetings with American diplomats, “China resolutely opposes hacking actions.”
In fact, however, the Chinese had been hacking, with growing profligacy, for more than a decade. A senior U.S. intelligence official had once muttered at an NSC meeting that at least the Russians tried to keep their cyber activity secret; the Chinese just did it everywhere, out in the open, as if they didn’t care whether anyone noticed.
As early as 2001, in an operation that American intelligence agencies dubbed Titan Rain, China’s cyber warriors hacked into the networks of several Western military commands, government agencies, defense corporations, and research labs, using techniques reminiscent of the Russians’ Moonlight Maze operation.
Around the same time, the Third Department of the PLA’s General Staff, which later created Unit 61398, adopted a new doctrine that it called “information confrontation.” Departments of “information-security research” were set up in more than fifty Chinese universities. By the end of the decade, the Chinese army started to incorporate cyber tools and techniques in exercises with names like “Iron Fist” and “Mission Attack”; one scenario had the PLA hacking into U.S. Navy and Air Force command-control networks in an attempt to impede their response to an occupation of Taiwan.
In short, the Chinese were emulating the American doctrine of “information warfare”—illustrating, once more, the lesson learned by many who found the cyber arts at first alluring, then alarming: what we could do to an adversary, an adversary could do to us.
There was one big difference in the Chinese cyber attacks: they were engaging not just in espionage and battlefield preparation, but also in the theft of trade secrets, intellectual property, and cash.
In 2006, if not sooner, various cyber bureaus of the Chinese military started hacking into a vast range of enterprises worldwide. The campaign began with a series of raids on defense contractors, notably a massive hack of Lockheed Martin, where China stole tens of millions of documents on the company’s F-35 Joint Strike Fighter aircraft. None of the files were classified, but they contained data and blueprints on cockpit design, maintenance procedures, stealth technology, and other matters that could help the Chinese counter the plane in battle or, meanwhile, build their own F-35 knockoff (which they eventually did).
Colonel Gregory Rattray, a group commander in the Air Force Information Warfare Center (which had recently changed its name to the Air Force Information Operations Center), was particularly disturbed: not only by the scale of China’s cyber raids but also by the passivity of American corporations. Rattray was an old hand in the field: he had written his doctoral dissertation on information warfare at the Fletcher School of Law and Diplomacy, worked on Richard Clarke’s staff in the early years of George W. Bush’s presidency, then, after Clarke resigned, stayed on as the White House director of cyber security.
In April 2007, Rattray summoned several executives from the largest U.S. defense contractors and informed them that they were living in a new world. The intelligence estimates that pinned the cyber attacks on China were highly classified; so, for one of his briefing slides, Rattray coined a term to describe the hacker’s actions: “APT”—for advanced persistent threat. Its meaning was literal: the hacker was using sophisticated techniques; he was looking for specific information; and he was staying inside the system as long as necessary—weeks, even months—to find it. (The term caught on; six years later, Kevin Mandia titled his report APT1.)
The typical Chinese hack started off with a spear-phishing email to the target-company’s employees. If just one employee clicked the email’s attachment (and all it took was one), the computer would download a webpage crammed with malware, including a “Remote Access Trojan,” known in the trade as a RAT. The RAT opened a door, allowing the intruder to roam the network, acquire the privileges of a systems administrator, and extract all the data he wanted. They did this with economic enterprises of all kinds: banks, oil and gas pipelines, waterworks, health-care data managers—sometimes to steal secrets, sometimes to steal money, sometimes for motives that couldn’t be ascertained.
McAfee, the anti-virus firm that discovered and tracked the Chinese hacking operation, called it Operation Shady RAT. Over a five-year period ending in 2011, when McAfee briefed the White House and Congress on its findings, Shady RAT stole data from more than seventy entities—government agencies and private firms—in fourteen countries, including the United States, Canada, several nations in Europe, and more in Asia, including many targets in Taiwan but, tellingly, none in the People’s Republic of China.
President Obama didn’t need McAfee to tell him about China’s cyber spree; his intelligence agencies were filing similar reports. But the fact that a commercial anti-virus firm had tracked so much of the hacking, and released such a detailed report, made it hard to keep the issue locked up in the closet of diplomatic summits. The companies that were hacked would also have preferred to stay mum—no point upsetting customers and stockholders—but the word soon spread, and they reacted by pressuring the White House to do something, largely because, after all these decades of analyses and warnings, many of them still didn’t know what to do themselves.
This was the setting that forced Obama’s hand. After another Asia security summit, where his diplomats once again raised the issue and the Chinese once again denied involvement, he told Tom Donilon to deliver a speech that brought the issue out in the open. The Mandiant report—which had been published three weeks earlier—upped the pressure and accelerated the timetable, but the dynamics were already in motion.
One passage in Donilon’s speech worried some midlevel officials, especially in the Pentagon. Characterizing cyber offensive raids as a violation of universal principles, even as something close to a cause for war, Donilon declared, “The international community cannot afford to tolerate any such activity from any country.”
The Pentagon officials scratched their heads: “any such activity from any country”? The fact was, and everyone knew it, the United States engaged in this activity, too. Its targets were different: American intelligence agencies weren’t stealing foreign companies’ trade secrets or blueprints, much less their cash, mainly because they didn’t need to be; such secrets or blueprints wouldn’t have given American companies an advantage—they already had the advantage.
In NSC meetings on the topic, White House aides argued that this distinction was important: espionage for national security was an ancient, acceptable practice; but if the Chinese wanted to join the international economy, they had to respect the rights of property, including intellectual property. But other officials at these meetings wondered if there really was a difference. The NSA was hacking into Chinese networks to help defeat them in a war; China was hacking into American networks mainly to help enrich its economy. What made one form of hacking permissible and the other form intolerable?
Even if the White House aides had a point (and the Pentagon officials granted that they did), wasn’t the administration skirting danger by going public with this criticism? Wouldn’t it be too easy for the Chinese to release their own records, revealing that we were hacking them, too, and thus accuse us of hypocrisy? Part of what we were doing was defensive: penetrating their networks in order to follow them penetrating our networks; and we were penetrating these networks so deeply that, whenever the Chinese tried to hack into Defense Department systems (or, lately, those of several weapons contractors, too), the NSA was monitoring every step they took—it was monitoring what the Chinese were seeing on their own monitors. On a few occasions, the manufacturing secrets that the Chinese stole weren’t real secrets at all; they were phony blueprints that the NSA had planted on certain sites as honey pots. But, to some extent, these cyber operations were offensive in nature: the United States was penetrating Chinese networks to prepare for battle, to exploit weaknesses and exert leverage, just as the Chinese were doing—just as every major power had always done in various realms of warfare.
The whole business of calling out China for hacking was particularly awkward, given the recent revelations about Stuxnet, to say nothing of Obama’s recent (though still highly classified) signing of PPD-20, the presidential directive on cyber operations. Some of Obama’s White House aides acknowledged a certain irony in the situation; it was one reason the administration refused to acknowledge having played a role in Stuxnet, long after the operation had been blown.
In May, Donilon flew to Beijing to make arrangements for a summit between President Obama and his Chinese counterpart, Xi Jinping. Donilon made it clear that cyber would be on the agenda and that, if necessary, Obama would let Xi in on just how much U.S. intelligence knew about Chinese practices. The summit was scheduled to take place in Rancho Mirage, California, at the estate of the late media tycoon Walter Annenberg, on Friday and Saturday, June 7 and 8, 2013.
On June 6, The Washington Post and The Guardian of London reported, in huge front-page stories, that, in a highly classified program known as PRISM, the NSA and Britain’s GCHQ had long been mining data from nine Internet companies, usually under secret court orders and that, through this and other programs, the NSA was collecting telephone records of millions of American citizens. These were the first of many stories, published over the next several months by The Guardian, the Post, Der Spiegel, and eventually others, based on a massive trove of beyond-top-secret documents that NSA systems administrator Edward Snowden had swiped off his computer at the agency’s facility in Oahu, Hawaii, and leaked to three journalists before fleeing to Hong Kong, where he met with two of them, Laura Poitras and Glenn Greenwald. (The other reporter, Barton Gellman, couldn’t make the trip.)
The timing of the leak, coming on the eve of the Obama-Xi summit, was almost certainly happenstance—Snowden had been in touch with the reporters for months—but the effect was devastating. Obama brought up Chinese cyber theft; Xi took out a copy of The Guardian. From that point on, the Chinese retort to all American accusations on the subject shifted from “We don’t do hacking” to “You do it a lot more than we do.”
One week after the failed summit, as if to bolster Xi’s position, Snowden—who, by this time, had revealed himself as the source in a dramatic video taped by Poitras in his hotel room—said, in an interview with Hong Kong’s top newspaper, the South China Morning Post, that the NSA had launched more than 61,000 cyber operations, including attacks on hundreds of computers in Hong Kong and mainland China.
The Morning Post interview set off suspicions about Snowden’s motives: he was no longer just blowing the whistle on NSA domestic surveillance; he was also blowing foreign intelligence operations. Soon came newspaper stories about NSA hacking into email traffic and mobile phone calls of Taliban insurgents on the eastern border of Afghanistan; an operation to gauge the loyalties of CIA recruits in Pakistan; email intercepts to assist intelligence assessments of events in Iran; and a surveillance program of cell phone calls “worldwide,” intended to find and track associates of known terrorists.
One leak was the full, fifty-page catalogue of tools and techniques used by the elite hackers in the NSA’s Office of Tailored Access Operations. No American or British newspaper published that document, though Der Spiegel did, in its print and online editions. Fort Meade’s crown jewels were now scattered all over the global street, for interested parties everywhere to pick up. Even the material that no one published—and Snowden’s cache amounted to tens of thousands of highly classified documents—could have been perused by any foreign intelligence agency with skilled cyber units. If the NSA and its Russian, Chinese, Iranian, French, and Israeli variants could hack into one another’s computers, they could certainly hack into the computers of journalists, some of whom were less careful than others in guarding the cache. Once Snowden took his laptops out of the building in Oahu, its contents—encrypted or otherwise—were up for grabs.
But the leaks about foreign intelligence operations—the intercepts of email in Afghanistan and Pakistan, the TAO catalogue, and the like—were overshadowed, among American news readers, by the detailed accounts of domestic surveillance. It was these leaks that earned Snowden applause as a whistleblower and engulfed the NSA in a storm of controversy and protest unseen since the Church Committee hearings of the 1970s.
The Snowden papers unveiled a massive data-mining operation, more vast than any outsider had imagined. In effect, it was Keith Alexander’s metadata experiment at Fort Belvoir writ very large—the realization of his philosophy about big data: collect and store everything, so that you can go back and search for patterns and clues of an imminent attack; when you’re looking for a needle in a haystack, you need the whole haystack.
Under the surveillance system described in the Snowden documents, when the NSA found someone in contact with foreign terrorists, its analysts could go back and look at every phone number the suspect had called (and every number that had called the suspect) for the previous five years. The retrieval of all those associated numbers was called the first “hop.” To widen the probe, analysts could then look at all the numbers that those people had called (the second hop) and, in a third hop, the numbers that those people had called.
The math suggested, at least potentially, a staggering level of surveillance. Imagine someone who had dialed the number of a known al Qaeda member, and assume that this person had phoned 100 other people over the previous five years. That would mean the NSA could start tracking not only the suspect’s calls but also the calls of those 100 other people. If each of those people also called 100 people, the NSA—in the second hop—could track their calls, too, and that would put (100 times 100) 10,000 people on the agency’s screen. In the third hop, the analysts could trace the calls of those 10,000 people and the calls that they had made—or (10,000 times 100) 1 million people.
In other words, the active surveillance of a single terrorist suspect could put a million people, possibly a million Americans, under the agency’s watch. The revelation came as a shock, even to those who otherwise had few qualms about the occasional breach of personal privacy.
Following this disclosure, Keith Alexander gave several speeches and interviews, in which he emphasized that the NSA did not examine the contents of those calls or the names of the callers (that information was systematically excluded from the database) but rather only the metadata: the traffic patterns—which phone numbers called which other phone numbers—along with the dates, times, and durations of those calls.
But amid the dramatic news stories, an assurance from the director of the National Security Agency struck a weak chord: he might say that his agency didn’t listen to these phone calls, but many wondered why they should believe him.
The distrust deepened when Obama’s director of national intelligence, James Clapper, a retired Air Force lieutenant general and a veteran of various spy agencies, was caught in a lie. Back on March 12, three months before anyone had heard of metadata, PRISM, or Edward Snowden, Clapper testified at a public hearing of the Senate Select Committee on Intelligence. At one point, Senator Ron Wyden, Democrat from Oregon, asked him, “Does the NSA collect any type of data at all on millions or hundreds of millions of Americans?”
Clapper replied, “No, sir . . . not wittingly.”
As a member of the select committee, Wyden had been read in on the NSA metadata program, so he knew that Clapper wasn’t telling the truth. The day before, he’d given Clapper’s office a heads-up on the question that he planned to ask. He knew that he’d be putting Clapper in a box: the correct answer to his question was “Yes,” but Clapper would have a hard time saying so without making headlines; so Wyden wanted to give the director a chance to formulate an answer that addressed the issue without revealing too much. He was surprised that Clapper dealt with it by simply lying. After the hearing, Wyden had an aide approach Clapper to ask if he’d like to revise and extend his reply for the record; again to Wyden’s surprise, Clapper declined. Wyden couldn’t say anything further in public without violating his own pledge to keep quiet about high-level secrets, so he let the matter rest.
Then came the Snowden revelations, which prompted many to reexamine that exchange. On June 9, the first Sunday after the Snowden leaks were published, Clapper agreed to an interview with NBC-TV’s Andrea Mitchell. She asked him why he’d answered Wyden’s question the way he did.
Clapper came off as astonishingly unprepared. “I thought, though in retrospect, I was asked ‘when are you going to . . . stop beating your wife’ kind of question, which is . . . not answered necessarily by a simple yes or no,” he began in an incoherent ramble. Then, digging himself still deeper in the hole, he said, “So, I responded in what I thought was the most truthful—or least untruthful—manner by saying, ‘No.’ ”
Doubling down, Clapper homed in on Wyden’s use of the word “collect,” as in, “Did the NSA collect any type of data . . . on millions of Americans?” Imagine, Clapper said, a vast library of books containing vast amounts of data on every American. “To me,” he went on, “collection of U.S. persons’ data would mean taking the book off the shelf and opening it up and reading it.” Therefore, he reasoned, it wasn’t quite a lie to say that the NSA did not collect data on Americans, at least not wittingly.
The morning after the broadcast, Clapper called his old friend Ken Minihan, the former NSA director, to ask how he did. Minihan was now managing director of the Paladin Capital Group, which invested in cyber security technology firms worldwide: he’d been out of government for more than a decade, but he kept up his contacts throughout the intelligence world; he still had both feet in the game, and he’d watched Clapper’s interview in sorrow.
“Well,” Minihan replied, in his folksy drawl, “you couldn’t have made things any worse.”
Clapper might have been genuinely perplexed. Five years earlier, the FISA Court had allowed the NSA to redefine “collection” in exactly the way Clapper had done on national television—as the retrieval of data that had already been scooped up and stored. At the time of that ruling, Alexander was laying the foundations of his metadata program; it was illegal to “collect” data from Americans, so the program couldn’t have gone forward without redefining the term.
But the FISA Court was a secret body: it met in secret; its cases were heard in secret; its rulings were classified Top Secret. To Clapper and other veterans of the intelligence community, this reworking of a common English word had insinuated its way into official parlance. To anyone outside the walls, the logic seemed disingenuous at best. Clearly, to collect meant to gather, to sweep up, to bring together. No one would say, “I’m going to collect The Great Gatsby from my bookshelf and read it,” nor did it seem plausible that anyone in the NSA would say, “I’m going to collect this phone conversation from my archive and insert it in my database.”
The NSA had basked in total secrecy for so long, from the moment of its inception, that its denizens tended to lose touch with the outside world. In part, its isolation was a product of its mandate: making and breaking codes in the interest of national security ranked among the most sensitive and secretive tasks in all government. Yet the insularity left them without defenses when the bubble was suddenly pierced. They’d had no training or experience in dealing with the public. And as the secrets in Snowden’s documents were splashed on front-page headlines and cable newscasts day after jaw-dropping day, the trust in the nation’s largest, most intrusive intelligence agency—a trust that had never been more than tenuous—began to crumble.
Opinion polls weren’t the only place where the agency took a beating. The lashes were also felt, and more damagingly so, in the agitated statements and angry phone calls from corporate America—in particular, the telecoms and Internet providers, whose networks and servers the NSA had been piggybacking for years, in some cases decades.
This arrangement had been, for many firms, mutually beneficial. As recently as 2009, after the Chinese launched a major cyber attack against Google, stealing the firm’s source-code software, the crown jewels of any Internet company, the NSA’s Information Assurance Directorate helped repair the damage. One year earlier, after the U.S. Air Force rejected Microsoft’s XP operating system on the grounds that it was riddled with security flaws, the directorate helped the firm design XP Service Pack 3, one of the company’s most successful systems, which Air Force technicians (and many consumers) deemed secure straight out of the box.
Yet now with their complicity laid bare for all to see, the executives of these corporations backed away, some howling in protest, like Captain Renault, the Vichy official in the film Casablanca who pronounced himself “shocked, shocked to find that gambling is going on in here,” just as the croupier delivered his winnings for the night. Their fear was that customers in the global marketplace would stop buying their software, suspecting that it was riddled with back doors for NSA intrusion. As Howard Charney, senior vice president of Cisco, a company that had done frequent business with the NSA, told one journalist, the Snowden revelations were “besmirching the reputation of companies of U.S. origin around the world.”
Allied governments around the world were clamoring as well. The English-speaking nations that had been sharing intelligence with the United States for decades—the fellow “five-eyes” countries, Great Britain, Canada, Australia, and New Zealand—held firm. But other state leaders, who had not been let into the club, started slipping away. President Obama had planned to rally European leaders in his pressure campaign against China—which had launched cyber attacks on a lot of their companies, too—but his hopes were dashed when a Snowden document revealed that the NSA had once hacked German chancellor Angela Merkel’s cell phone. Merkel was outraged.
There was more than a trace of Captain Renault in her fuming, too; as subsequent news stories revealed, the BND, Germany’s security service, continued to cooperate with the NSA in monitoring suspected terrorist groups. But at the time, Merkel played populist, as vast swaths of the German people, including many who had once seen America as a protector and friend, started likening the NSA to Stasi, the extremely intrusive surveillance service of the long-imploded East German dictatorship. Other Snowden documents exposed NSA intercepts in Central and South America, infuriating leaders and citizens in the Western Hemisphere, as well.
Something had to be done; the stench—political, economic, and diplomatic—had to be contained. So President Obama did what many of his predecessors had done in the face of crises: he appointed a blue-ribbon commission.