The question now is whether Congress can summon the will to hold the president to account.
In 1975, Congress established the first intelligence oversight committees. The Watergate break-in had prompted a constitutional crisis, and in the course of investigating Watergate, Congress realized that it had an urgent need to review classified documents and activities in order to serve as a check on a president who was willing to use the nation’s intelligence agencies to investigate his political rivals, to manipulate popular opinion, and—ironically—to assess whether the Russian government was fomenting civil and political unrest in the United States.
A president who is willing to bend national policy to serve his self-interests, and who has little regard for the law, has powerful tools of concealment at his disposal. The events of the past two weeks are again putting on full display the ways that classification policy and information-security systems can be abused to conceal records of wrongdoing and to extend the life span of malfeasance before it’s discovered. And once again, a great deal rides on whether Congress can summon the will to serve as a check on the executive.
In late September, the House Permanent Select Committee on Intelligence released an unclassified complaint filed by an intelligence-community whistle-blower alleging that President Donald Trump was using his office to pressure foreign governments to interfere in U.S. elections. The charges themselves were explosive, and seem to have been confirmed in subsequent days, as Trump has publicly urged Ukraine and China—and by extension, every other world government—to dig up dirt on his political rivals. Yet it wasn’t just the top-level charges, as impeachment-worthy as they are, that caught the attention of former national-security officials. It was also the allegation that someone in the White House had moved call transcripts into a highly restricted server where records of Trump’s discussions with foreign leaders would be concealed.
With Trump’s public admissions, and the White House’s release of summary notes of a July call between Trump and Ukrainian President Volodymyr Zelensky, why should anyone care what server is used to store the call notes? The answer lies in the nature of classification decisions, and the ways in which this administration appears to be using classification controls to conceal evidence of crimes and wrongdoing. I spent nearly 15 years working with classified information, classification policy, and intelligence oversight, first at the National Security Agency, where I served in a number of roles, including as associate general counsel for intelligence law, and later as senior minority counsel for the Russia investigation on the Senate Select Committee for Intelligence. To be clear, I don’t speak for these agencies, and I haven’t had any firsthand exposure to the call memos at issue here—but I do have some understanding of how classification works, and why these allegations about the handling of the call transcripts matter.
The allegations in the whistle-blower’s complaint were so shocking that they might have been dismissed out of hand had they not been written with such precision and care, and had they not borne such a near-perfect correspondence with the White House’s own notes of the Trump-Zelensky call. Within a week of the complaint’s release, new allegations of concealment emerged, indicating that the White House had used this same server to bury notes from the Oval Office meeting between Trump and Russian Foreign Minister Sergey Lavrov days after Trump fired FBI Director James Comey; the contents of calls between Trump and the Saudi royal family in the aftermath of the Saudi government’s murder of the Saudi journalist Jamal Khashoggi; and Trump’s calls with Chinese President Xi Jinping. The pattern of enlisting help from foreign governments for partisan domestic purposes was most starkly revealed by Trump’s own public admissions on the White House lawn, as he urged Ukraine and China to investigate the front-runners in the Democratic-primary race. But if Trump had been cagier, if he had kept his cards closer to the vest, the fact that those call transcripts were placed off-limits might have meant it would take decades, or longer, for this betrayal of national interests to come to light. And legitimate government tools—classification markings and restricted systems—would have been co-opted into concealing the crimes.
In order to understand why the use of restricted systems to carry out de facto overclassification of information matters, it helps to understand how classification decisions work, and what effect those decisions have.
National-security classifications—the designations of CONFIDENTIAL, SECRET, and TOP SECRET—can be slippery things. Some matters of national importance—weapons schematics, troop deployments, the identity of intelligence assets—should be kept off-limits, accessible only to a small number of individuals with authorized roles in government service who need that information to do their job. However, deciding the specific level of classification to apply is often as much a matter of art as science; there’s no mathematical formula to apply. In the absence of concretely quantifiable measures, “original classification authorities” (OCAs)—the handful of individuals authorized to make unilateral, or “original,” classification decisions about government information—are supposed to rely on the guidance in Executive Order 13526.
Signed in 2009, EO 13526 explains that the three classification levels indicate whether unauthorized access to particular information would cause “damage,” “serious damage,” or “exceptionally grave damage” to the interests of the United States. There is also an entire set of additional labels—handling caveats, restrictive markings, dissemination controls, and compartments—that agencies can use to impose fine-grained controls on information of all kinds, so that information can be viewed only by a subset of the individuals who hold a particular level of clearance. Because the executive order requires OCAs to be able to articulate what that damage would be, and because there are only a few dozen OCAs but tens of thousands of subordinate employees who must apply appropriate classification markings on a daily basis, agencies have developed the practice of creating comprehensive classification guides that describe what markings should be applied to which information and under what circumstances. For example, the statement that “the U.S. IC gathers intelligence on international terrorist organizations” is unclassified. As more information is added—which IC agencies, which specific terrorist organizations, in what parts of the world, via what intelligence-gathering sources or methods, etc.—the addition of these details eventually causes the statement to become classified, because either some particular facet of those details (such as the particular techniques used to gather the intelligence) or the combination of details are sufficiently perishable, fragile, or sensitive that revealing them publicly could cause damage, serious damage, or exceptionally grave damage to U.S. national-security interests.
EO 13526’s broadly phrased guidance has, predictably, led to classification decisions varying from one agency to another. And it isn’t just information sensitivity that drives classification decisions; they can also be shaped by agency culture and information technology. For example, for years most of the FBI’s computer systems were rated to handle only SECRET-level information; as a result, the FBI developed a culture of classifying information at the SECRET level. These cultural and IT-shaped preferences can carry on for years, and were on display in a batch of recently declassified documents that included an NSA affidavit, classified TS//SCI, and an FBI affidavit, classified SECRET//NOFORN. Both documents were created in the same year and served a similar purpose for the same surveillance program, yet they carried two very different classification levels.
When it comes to the mechanics of information protection, there are clearly defined standards for what kinds of systems can be used to hold varying levels of classified information. Some systems can apply different protections to individual pieces of information based on their level of classification. Other systems operate at “system-high,” treating all information as if it were subject to the highest level of protection that governs the most highly classified records on the system. The systems that hold the most sensitive, most tightly restricted information are often completely separated—“air-gapped”—from other systems, and physical and digital access to them is strictly limited to individuals who not only have the highest levels of security clearance but also have been “read into” the most tightly restricted programs in the federal government.
The protections don’t rest solely on document markings and IT-system permissions; respect for classification levels is deeply ingrained in clearance holders. Not only do they protect information they’ve been cleared for; by and large, they wouldn’t dream of trying to gain access to systems or data for which they hadn’t yet been properly cleared. As a general matter, the classification system works well to impose a systematic approach to describing the level of sensitivity of particular pieces of information, as well as creating a shorthand denotation of the combination of personnel clearance, facility clearance, and IT-system security needed to protect that information. The overall result is a strong culture of compliance with classification policy, and a set of procedures and systems that are designed to support the tightest imaginable controls over information that should, in fact, be tightly controlled.
As the whistle-blower’s complaint suggests, the very features that make the classification system work can also be exploited by high-level officials who are willing to violate norms, or by staff members who are willing to break the law on their behalf. What better place to hide the bodies than a burying ground that’s off-limits to almost everyone in government—and where no one is likely to even try to break in?
EO 13526 makes clear that classification can never be used as the basis for concealing evidence of wrongdoing or embarrassing information. In this case, document markings—the actual classification labels—weren’t changed. But there has been a de facto increase in classification, because the information is now held in a system that can be accessed only by a vanishingly small number of inner-circle White House officials.
It would be typical for notes from a diplomatic call to be classified at the SECRET//NOFORN level—as indicated by the classification markings on the Trump-Zelensky call summary that was released by the White House. If this document, or an even more detailed version of it (a true word-by-word transcript, rather than the released notes, which appear to be what analysts would refer to as a gist), was classified SECRET//NOFORN but then placed into a highly classified system, that would have the same practical effect as upgrading the classification—that is, of restricting the call notes at the same level as more highly classified information. If the notes were moved for the purpose of concealing wrongdoing or preventing embarrassment, those actions would almost certainly violate EO 13526.
It’s unlikely that the president can be charged with breaking any laws for the mere fact of relocating information. After all, he’s an OCA who can make classification determinations. Perhaps more important, as the president, he can amend or rescind an executive order with the stroke of a pen. However, the use of these tools to conceal wrongdoing could be charged as an article of impeachment. And his subordinates don’t enjoy the president’s authorities or protections. If the multi-sourced allegations being reported in recent weeks are true, someone moved the call transcripts from one system to the other. And it’s a safe bet that the president didn’t personally carry that out.
HPSCI Chairman Adam Schiff faces a tough challenge ahead. Senator Richard Burr, the chair of the Senate Select Committee on Intelligence, will as well, if he embraces the role of institutionalist and asserts the Senate’s authority to serve as a check on the executive branch. Whether HPSCI and SSCI succeed in obtaining information from the White House is crucial to the future of our nation—not just because the president’s public statements urging foreign leaders to investigate his political rivals are so shocking that they merit the highest-priority investigation, but also because the overall integrity of our framework of government, as well as the future effectiveness of our intelligence and national-security apparatus, demand it.
If Congress can’t obtain classified information from the executive branch, our much-vaunted system of checks and balances threatens to collapse. A failure by Congress to exercise oversight of classified matters risks ongoing harm: to transparency, to the nation’s ability to control and oversee activities that necessarily have to remain out of sight, and to public confidence in our national-security apparatus. Who would have thought the fate of the nation could rest on a question of which set of motherboards stored the memory of a particular batch of electrons, or on arcane questions of classification policy?
In its draft report, the 1975 House Intelligence Committee wrote that:
The key to exercising oversight is knowledge.
The Select Committee’s most important work may well have been its … testing perhaps for the first time what happens when Congress unilaterally decides what it wants to know and how it wants to know it.
Congress is faced squarely with that choice once again. How—and how hard—it chooses to pursue the information it’s entitled to may determine the future of Trump’s presidency, and of Congress’s future ability to oversee the executive branch.