@War: The Rise of the Military-Internet Complex

Home > Other > @War: The Rise of the Military-Internet Complex > Page 20
@War: The Rise of the Military-Internet Complex Page 20

by Shane Harris

As Alexander and his lieutenants saw things, it would be irresponsible bordering on negligent not to assist the department however they could. But that didn’t mean surrendering NSA’s role as the center of gravity in cyber security. The agency was part of the Department of Defense, and its writ extended to protecting the nation from foreign attacks, whether on land, in the air, at sea, or on a computer network.

  Chertoff and Alexander got along well, according to former officials who worked with both of them, and the secretary seemed happy to let the cyber warriors at Fort Meade take the lead. Alexander spent the next four years building up NSA’s cyber forces, culminating in the successful Buckshot Yankee operation and the establishment of Cyber Command. In 2009, Obama named former Arizona governor Janet Napolitano as his homeland security secretary. Alexander told his staff to give Napolitano and her team whatever help and advice they needed. But he had no intention of ceding the battlefield. Not when he was about to launch his biggest campaign yet.

  Alexander had seen how the Defense Industrial Base Initiative was able to give the government access to information from corporate computer networks. The companies had become digital scouts in cyberspace, and the information they reported back helped to feed NSA’s catalog of threat signatures—the lists of known malware, hacker techniques, and suspect Internet addresses that could be used to fortify defenses. Alexander liked to call it “the secret sauce.” The DIB had started with just twenty companies. Now he wanted to use the DIB model in new industries, including the energy and financial sector, and to bring as many as five hundred companies into the fold.

  At NSA the plan became known as Tranche 2. Operators of “critical infrastructure”—which could be broadly defined to include electrical companies, nuclear power plant operators, banks, software manufacturers, transportation and logistics companies, even hospitals and medical device suppliers, whose equipment could be hacked remotely—would be required by law or regulation to submit the traffic to and from their networks for scanning by an Internet service provider. The provider would use the signatures supplied by the NSA to look for malware or signs of a cyber campaign by a foreign government. It was a version of Alexander’s original plan to make the NSA the central clearinghouse for cyber threat intelligence. The NSA wouldn’t do the scanning, but it would give all the requisite threat signatures to the scanner. That helped the NSA avoid the impression that it was horning its way into private computer networks, even though it was actually in control of the whole operation. Once the scanners detected a threat, NSA analysts would move in and assess it. They would decide whether to let the traffic pass or to block it, or, if need be, to strike back at the source.

  The agency had already developed a scanning system called Tutelage that could isolate e-mails containing viruses and put them in a kind of digital petri dish, so that analysts could examine them without infecting any computers. This was the “sensor, sentry, and sharpshooter” that the NSA had used to monitor its Internet gateways back in 2009. Now Alexander wanted to bring that capability to bear as part of Tranche 2, effectively turning hundreds of companies and critical- infrastructure operators into a new front in the cyber wars.

  This made some Obama administration officials nervous. The president had clearly stated his intentions to protect cyberspace as a critical national asset. But he had always been conflicted about how long a leash to give the NSA. Obama had never warmed to the agency or Alexander. And although he appreciated and embraced the powerful capabilities that the NSA had to offer, the culture of espionage seemed alien to him.

  In the summer 2009, Pentagon officials drafted an “execute order” that would allow the military to launch a counterstrike on computers sending malicious traffic not just to a military system but also against privately owned critical-infrastructure facilities, such as electrical power stations. That was an extraordinary step. Heretofore, the government had only given assistance to companies in the form of intelligence about hackers and malware, which they could use to bolster their own defenses. Now the NSA wanted authority to launch a defensive strike against anyone attacking key American businesses in such a way that loss of life might occur—a blackout, say, or an attack on the air traffic control system—or if the US economy or national security would be jeopardized. That latter set of criteria was arguably broad and open to interpretation. Would a massive denial of service attack against American banks, for instance, which didn’t shut them down or steal funds but disrupted their operations, count as a hostile act that jeopardized the US economy?

  Obama administration officials pared back the order—but only slightly. Obama didn’t push the NSA out of the business of retaliatory strikes. He just required it to get authorization from him or his secretary of defense.

  Perhaps sensing that he couldn’t always count on Obama’s unconditional support, Alexander took his plans for Tranche 2 to Capitol Hill and the lawmakers who controlled his agency’s multibillion-dollar budget. Alexander told them and their staff that he supported legally requiring companies to share their data with government-appointed traffic scanners. But that was not a proposal the administration supported, at least not in its current form. White House aides had to admonish Alexander several times during 2011 and 2012, when a cyber bill was moving through Congress, not to speak on the president’s behalf and make promises that the administration wasn’t sure it could keep.

  “They’re pretty mad at me downtown,” Alexander said sheepishly in one meeting with congressional staffers. But that didn’t stop him from pushing harder. Alexander was an awkward public speaker, but in small groups he could be charming and compelling. He formed alliances with the Democratic and Republican chairs of the House and Senate Intelligence Committees. Lawmakers gave him the money he wanted and appropriated new funds for cyber security. Congressional oversight of NSA’s activities was minimal and nonintrusive. Alexander was winning the war on Capitol Hill. But inside the administration, he had enemies.

  By the time she arrived at the Homeland Security Department as the new deputy secretary in early 2009, Jane Holl Lute found that a battle for control of cyber security had already been fought—and Alexander had won. Many of her colleagues had long since concluded that the NSA was the only game in town, because it was the only agency with an extensive catalog of threat signatures, including malware, hacker techniques, and suspect Internet addresses. They knew that information had been gleaned from classified, expensive intelligence-gathering operations, which gave it a certain cachet and credibility. They also knew that Homeland Security had no comparable store of information, and scarcely a cyber security staff to speak of. The department employed twenty-four computer scientists in 2009, while the Defense Department employed more than seven thousand, most of whom worked at the NSA. Homeland Security’s computer-emergency watch center also couldn’t monitor network traffic in real time, making it practically useless as an early-warning system for cyber attacks. The best Homeland Security could hope to do was play a public relations role, encouraging companies to adopt good “cyber hygiene,” better monitor their own networks, and share information with the government. But these were gestures, not actions.

  The first time Lute met the official in charge of the department’s budding cyber defense mission was when he handed her his letter of resignation. In March, Rod Beckstrom quit in protest over what he described as the NSA’s interference in policies that, by law, were Homeland Security’s responsibility. “NSA effectively controls DHS cyber efforts,” Beckstrom wrote in a scathing rebuke. The NSA had stationed its employees in the department’s headquarters and installed its own proprietary technology. And recently NSA leaders had proposed relocating Beckstrom and his staff—all five of them—to the agency’s headquarters at Fort Meade.

  “During my term as director, we have been unwilling to subjugate the [center] underneath the NSA,” Beckstrom wrote. He warned Lute, Napolitano, and the president’s top national security advisers, including Secretary of Defense Robert Gates, that if the NSA were given
the reins, it would run roughshod over privacy and civil liberties and subsume the department into a culture of secrecy.

  Lute was no cyber expert. A former army officer, she last served managing peacekeeping operations for the United Nations. But as the de facto chief operating offer of the department, she’d been charged with making sense of its muddied cyber policies. Clearly, that was going to entail battle with the NSA. (Napolitano didn’t want the job, and was arguably unqualified for it. Practically a technophobe, she had no personal online accounts, and even at work she didn’t use e-mail.)

  Lute had been around intelligence officials long enough to conclude that they gained much of their power from secrecy, and by cultivating an appearance of omniscience. She didn’t adhere to the conventional wisdom that only the NSA had the know-how to defend cyberspace. “Pretend the Manhattan phone book is the universe of malware,” she once told colleagues. “NSA only has about one page of that book.” Lute thought that many companies already had the most important threat signatures, because they were collecting them from the hackers and foreign governments who tried to break in to their networks every day. Private security companies, antivirus researchers, even journalists were collecting and analyzing malware and other threat signatures, and either selling the information or publishing it as a public service. Software companies sent out automatic patches to fix known holes in their programs. The NSA tracked all this information. Why should anyone presume their intelligence didn’t incorporate what was already widely known? The spy agency’s information might be helpful, but companies didn’t require it to defend themselves, Lute said. They needed to share what they knew with one another, like an Internet version of a neighborhood watch.

  Lute wasn’t alone in thinking that Alexander had oversold his “secret sauce.”

  “There’s a presumption that if something is classified, it must be true, which is not remotely the case,” says a senior law enforcement official who sparred with NSA officials in several meetings about whether it should take the leading role in defending companies’ computer networks. “We can lay out information to a policymaker that’s ‘law enforcement sensitive’ [a lower level of classification than top secret], and they’ll say, ‘No, we’ve got this top-secret report, it must be true.’ And that’s hard to refute, because the NSA doesn’t bring the facts to the table about how it got that information or whether it’s unique. Policymakers and the public are not getting an accurate picture of the threat.”

  Even when Alexander met with senior executives from the world’s biggest technology firms, including Google, who knew plenty about cyber spies and attackers and had a financial interest in stopping them, he tried to persuade them that the NSA’s intelligence was superior. “His attitude was, ‘If only you knew what we knew, you’d be very afraid. I’m the only one that can help you,’” says a former senior security official.

  “Alexander convinced many lawmakers and policymakers that the NSA had a monopoly on this and it was all at Fort Meade,” says the former administration official who worked on cyber security issues. “And he’d use that phrase, ‘secret sauce.’ I’ve been behind the curtain up there; there is no secret sauce. It’s complete bullshit.”

  A low-grade tension persisted for the first two years of Lute’s tenure at Homeland Security. In February of 2011, it erupted into a public turf war. At a defense industry conference in Colorado Springs, the home of the US Air Force Academy, Alexander declared that the NSA should take a leading role in protecting cyberspace, the fifth domain of warfare. He called for new powers to defend against potentially crippling attacks on the United States. “I do not have the authority to stop an attack against Wall Street or industry, and that’s a gap I need to fix,” he said. Alexander had thrown down the gauntlet, effectively declaring US cyberspace a militarized zone.

  Alexander was scheduled to give a version of the same talk eight days later at one of the biggest annual computer security conferences, in San Francisco. Major newspapers and technology trade press would be there. Lute cut him off at the pass. On February 14, three days ahead of his speech, she and another senior Homeland Security official published an online op-ed for Wired, the influential technology magazine. “These days, some observers are pounding out a persistent and mounting drumbeat of war, calling for preparing the battlefield, even saying that the United States is already fully into a ‘cyberwar,’ that it is, in fact, losing,” Lute wrote. “We disagree. Cyberspace is not a war zone.”

  It was a direct shot at Alexander. “Conflict and exploitation are present there, to be sure, but cyberspace is fundamentally a civilian space,” Lute wrote, “a neighborhood, a library, a marketplace, a school yard, a workshop—and a new, exciting age in human experience, exploration and development. Portions of it are part of America’s defense infrastructure, and these are properly protected by soldiers. But the vast majority of cyberspace is civilian space.”

  Alexander was undeterred. He gave his speech as scheduled and repeated the same themes. And a few days later he fired back at Lute. “There’s a lot of folks that say we’d like the technical capabilities of NSA . . . but we don’t want NSA in there” protecting networks, Alexander said at a conference in Washington about domestic security, which was the Homeland Security Department’s domain. He bristled at the suggestion that his agency should lean back and only help defend when asked, rather than rush to the front lines. Alexander even invoked the Maginot Line, the long stretch of concrete fortifications France built along its border with Germany in the 1930s, suggesting that the United States risked being overrun if it focused its defense purely on strategy and underestimated the cunning of their enemies. (The Nazis overcame the line by going around it, a move the French hadn’t planned for, and ultimately conquered the country in six weeks.)

  The turf war was getting hot. The White House ultimately nixed Alexander’s Tranche 2 plan, not because Obama thought the NSA wasn’t up to the job of defending cyberspace but because it looked too much like a big government-surveillance program. The administration didn’t abandon Alexander’s core idea. It opted instead to use the existing DIB program, which was itself a big government-surveillance program, to test whether Internet service providers could monitor traffic using classified government intelligence—that NSA secret sauce. It was a compromise. The NSA wouldn’t get access to companies’ networks, but it would funnel intelligence to them through the Internet service providers.

  In the spring of 2011, seventeen defense companies volunteered for the test. The NSA still gave threat signatures to three service providers—CenturyLink, AT&T, and Verizon. The latter two were intimately familiar with NSA surveillance, having been a part of the agency’s bulk collection of Americans’ phone records since shortly after the 9/11 terrorist attacks. And all three companies were accustomed to handing over e-mails and online data about their customers at the request of the FBI and NSA.

  The test focused on two specific countermeasures: quarantining incoming e-mails infected with malware and preventing outbound traffic from contacting malicious Internet addresses, a process known as sinkholing. Most organizations only monitored traffic coming into their networks and ignored data that was being sent from inside their systems. Hackers took advantage of that ignorance and frequently disguised a company’s own documents as legitimate outbound traffic, before sending it on to a server under the hackers’ control.

  The test was a qualified success. An independent review by Carnegie Mellon University, one of the top technology research institutions in the country, found that the Internet service providers were able to receive the classified threat signatures and keep them secret. But there was some bad news for the vaunted cyber warriors at Fort Meade: practically none of the signatures told the companies anything they didn’t already know, a finding that supported Lute and others who doubted the power of Alexander’s secret sauce.

  Most of NSA’s intelligence was out of date by the time it was received. Of fifty-two cases of malicious activity that were
detected during the test, only two were the result of NSA threat signatures. The rest the companies found on their own, because they’d spent the last few years building their own network-monitoring capabilities and beefing up their defenses.

  The NSA could take some pride in knowing that those companies got so much better at defense because of their early participation in the DIB program, back in 2007, when they’d been essentially required to hand over threat information and take the government’s help if they wanted to keep doing business with the military. But the pilot undercut Alexander’s argument that his agency was uniquely qualified to protect the nation.

  Not that the companies needed a university study to tell them that. As early as 2010, corporate executives began to question whether the NSA was as sophisticated as Alexander claimed. During a meeting with CEOs at Homeland Security Department headquarters, Alexander gave a presentation on the NSA’s threat signature catalog. According to one participant, Google CEO Eric Schmidt leaned over to the person sitting next to him and whispered, “You mean to tell me they spent all this money and this is what they came up with? We’ve all moved beyond this now.” Google, like many other large companies that were frequent targets of hackers, had its own sources of threat intelligence from private security companies—such as Endgame, which sells zero day information—and had begun its own intelligence-gathering operations on hackers in China. But the company was also using other tactics, such as implementing stronger encryption for its users, and moving toward a “secure sockets layer” service that would set end-to-end encryption by default for everyone logged in to their Google account. Threat signatures alone “don’t work anymore,” Schmidt said. “The threats don’t just come where the NSA points its sensors.” Hackers were constantly changing their techniques and looking for new points of entry. They knew that the government was monitoring them—that’s why they changed up their tactics.

 

‹ Prev