Spy and Tell

On October 25, 1962, at the height of the Cuban missile crisis, Adlai Stevenson, the U.S. ambassador to the UN, confronted his Soviet counterpart, Valerian Zorin, in the chamber of the Security Council. Live on television, Stevenson grilled Zorin about whether the Soviet Union had deployed nuclear-capable missiles to Cuba. “Yes or no?” Stevenson demanded. As Zorin waffled, Stevenson went in for the kill: “I am prepared to wait for an answer until hell freezes over if that’s your decision. And I’m also prepared to present the evidence in this room.” Stevenson then revealed poster-sized photographs taken by a high-altitude U-2 spy plane, images that showed Soviet missile bases in Cuba and directly contradicted Moscow’s denials. Stevenson’s revelations marked a turning point in the crisis, providing undeniable evidentiary support to the Kennedy administration’s allegations, shifting global opinion, and pressuring the Soviets to de-escalate by isolating them diplomatically. It was the first time the U.S. government had declassified top-secret intelligence to publicly refute another country’s claims.

Nearly 60 years later, Moscow looked poised to flex its muscle again, this time by amassing nearly 175,000 troops on the Russian border with Ukraine. Echoing the Kennedy administration’s approach, the Biden administration responded by publicly disclosing intelligence, both to warn allies (and Ukraine) of the coming invasion and to preemptively rebut Russian President Vladimir Putin’s planned pretexts for it. In early December 2021, administration officials started sharing the intelligence community’s growing concern with the media, holding a briefing that was accompanied by satellite imagery showing Russian forces staging on Ukraine’s borders. In mid-January 2022, John Kirby, then the Pentagon press secretary, told reporters that Russia was preparing a “false-flag operation” in eastern Ukraine, hoping to fabricate a massacre to justify an invasion. Later that month, U.S. officials revealed that the Russian military had moved blood supplies to the border of Ukraine, suggesting that war was imminent. And on February 18, President Joe Biden said he was “convinced” that Russia’s invasion would begin in the “coming days”—as it did.

The Biden administration’s disclosures didn’t persuade Putin to shelve his war plans, but they did fortify Western resolve after the invasion. Advance warning of Russia’s plans enabled many U.S. allies, particularly NATO members, to quickly offer military aid packages to Ukraine and harmonize their economic sanctions against Russia. The revelations about the contrived provocations that Putin was scheming helped turn public opinion in the West against Russia by denying him a pretext for the invasion. Inside the Biden administration, the disclosure strategy was seen as a resounding success.

In the six decades after the Kennedy administration’s novel move at the UN, successive White Houses adopted the tactic from time to time. But their disclosures were “one and done” affairs. What is new now is that the Biden team has disclosed information multiple times on a single issue over an extended period. What’s more, because the Ukraine-related revelations seemed to work so well, the administration is now applying the tool to other issues, most notably China. It has even come up with terms for the practice, with officials speaking of “strategic downgrades” and “strategic declassification.” What used to be a break-glass option is now routine.

But as strategic downgrades become more common, policymakers and intelligence practitioners need to develop guardrails to protect against their pitfalls. Without proper precautions, a disclosure might compromise the source of the declassified information or, if the revelation turns out to be wrong, harm the intelligence community’s reputation and undermine the goal the disclosure was meant to achieve. The biggest risk, however, is that using intelligence as a policy tool increases the chances that it will also be used as a political weapon. Were that to happen, the intelligence community could lose its most precious asset: its reputation for objectivity.

A NOT-SO-SECRET HISTORY

Although high-level officials have long leaked classified intelligence to the media, strategic disclosures are something different. They aim to use intelligence to further a specific administration goal rather than advance a particular bureaucratic player’s individual interest. Accordingly, disclosures are known in advance by a group of senior officials, including those with declassification authority, and are usually coordinated with relevant stakeholders, including the agency that collected the intelligence. They can enter the public domain directly—for example, through an on-the-record press conference, a televised speech or interview, or an intelligence product posted on a government website. Or they can take a more circuitous route, such as through a background briefing to journalists, who can use the information but agree not to name the official providing it. Strategic downgrades may or may not go through a formal declassification process, but unlike unauthorized leaks, they are legal, because officials with declassification authority have been involved in the decision-making.

Since the Cuban missile crisis, administrations have resorted to strategic declassification for a variety of reasons. Sometimes, the goal is to preemptively justify a policy. That was the purpose of the memorable, but ultimately incorrect, speech that Secretary of State Colin Powell gave to the UN Security Council in February 2003. Flanked by George Tenet, the director of the CIA, Powell played a tape of an intercepted conversation between Iraqi military officers conspiring to mislead weapons inspectors, showed satellite imagery of alleged weapons sites, and displayed drawings of supposed biological weapons labs. President Barack Obama made a similar move in 2013, after the Syrian regime of Bashar al-Assad fired rockets filled with sarin gas at a Damascus suburb and killed more than a thousand civilians. As the White House contemplated airstrikes, it released a summary of the intelligence community’s “high-confidence assessment” that the Syrian government had carried out the attack. In the end, the administration decided not to respond militarily, but had it done so, the declassified intelligence would have been foundational to the case for action by contradicting Syria’s repeated denials of responsibility.

At other times, disclosures are made to retroactively justify a policy. Such was the case in 1983, when Soviet pilots shot down a South Korean commercial airliner that had strayed into Soviet airspace. U.S. President Ronald Reagan declassified signals intelligence to show Soviet culpability and justify his confrontational posture toward Moscow. At the Security Council, the U.S. ambassador to the UN, Jeane Kirkpatrick, played a tape of Soviet pilots’ intercepted radio conversations with their commander as they homed in on the plane. Three years later, the administration repeated the strategy with Libya. After ordering airstrikes against the regime of Muammar al-Qaddafi for having orchestrated a terrorist attack that killed U.S. troops at a discotheque in West Berlin, Reagan, in a speech from the Oval Office, summarized diplomatic cables intercepted by the National Security Agency that proved Libyan responsibility for the attack.

Sometimes, policymakers disclose intelligence to undermine or pressure their adversaries. In 1984, the Reagan administration released declassified sketches based on classified satellite photography showing that the Soviets were constructing a radar station in Siberia, an outpost the administration claimed violated the Anti-Ballistic Missile Treaty. The goal was to strengthen the United States’ position in arms control talks and demonstrate that the U.S. government was closely monitoring Soviet military developments. (Years later, the Soviets dismantled the facility.) More recently, in 2009, Obama held a press conference with his British and French counterparts and announced that the Iranians had built a covert uranium enrichment site. As an administration official explained to reporters in an accompanying background briefing, the conclusion was based on “very sensitive intelligence information.” The disclosure worked: it generated international pressure on the Iranians, compelling them to bring the site under International Atomic Energy Agency safeguards.

At other times, the U.S. government has used official disclosures to deal with unauthorized ones. In 2007, the George W. Bush administration worried that an intelligence estimate about Iran’s nuclear program would leak. The estimate concluded that Iran had halted its nuclear weapons program four years earlier, and the White House feared that the revelation of that specific conclusion would undermine its argument that Iran still posed a threat. So it released an unclassified version of the paper’s key judgments to make clear that the country was continuing to work on both uranium enrichment and dual-use weapons technologies. The Obama administration resorted to the same strategy in 2013. When the National Security Agency contractor Edward Snowden leaked highly classified documents about the U.S. government’s global surveillance programs, the administration responded with its own disclosures. It released an overview of the programs to counter media stories that characterized them as being more pervasive and less subject to legal scrutiny than they actually were. The NSA’s director, Keith Alexander, even made an unprecedented appearance on 60 Minutes to share several previously classified pieces of information (such as the fact that the NSA was targeting the communications of fewer than 60 “U.S. persons” worldwide).

Other disclosures are motivated by an administration’s desire to protect its reputation. In 2004, a member of the 9/11 Commission questioned National Security Adviser Condoleezza Rice about a seemingly damning item that appeared in the President’s Daily Brief on August 6, 2001: “Bin Ladin Determined To Strike in US.” Just two days later, the Bush administration released the top-secret memo with minimal redactions to show that the document contained no specific warning of, or any actionable information about, a near-term attack. Likewise, in 2016, when the Obama administration wanted to counter criticism about civilian casualties caused by U.S. drone strikes, it released the intelligence community’s own count, which was much lower than the number calculated by critics.

Finally, U.S. policymakers have at times released intelligence to pressure Congress. In December 2023, Section 702 of the Foreign Intelligence Surveillance Act was set to expire. The provision allows the U.S. government to access the communications of foreigners outside the United States who have been targeted for intelligence purposes and whose communications pass through the United States. To encourage Congress to reauthorize Section 702, the administration declassified information showing its value. Officials revealed that the provision had proved crucial to tracking fentanyl smuggling across the Mexican border and identifying the hacker behind the 2021 ransomware attack on Colonial Pipeline. The administration even disclosed that in 2022, 59 percent of the pieces in Biden’s President’s Daily Briefs contained information collected under the authorities of Section 702. But the disclosures don’t appear to have worked: although the program was temporarily reauthorized, as of this writing, in March, permanent reauthorization remains stalled in Congress.

WEAPON OF CHOICE

The Biden administration’s disclosures about Russia’s war in Ukraine did not stop when the shooting started. In fact, they only gained pace. A month after the invasion, Biden revealed that Russia was considering using chemical and biological weapons in the conflict. By the end of 2023, with domestic enthusiasm for continued support for Ukraine flagging and Congress at an impasse over aid, it yet again resorted to strategic declassification. To demonstrate Ukraine’s success in the war and the effectiveness of U.S. military aid, it released the U.S. intelligence community’s estimate that Russia had suffered an astonishing 315,000 casualties since the invasion.

The Biden administration is now using strategic downgrades against China, too. In August 2022, on the eve of Speaker of the House Nancy Pelosi’s trip to Taiwan, Kirby, by then a National Security Council spokesperson, shared details from a declassified assessment of actions Beijing could take to register its displeasure, such as launching missiles into the Taiwan Strait. The goal: to preemptively remove the sting from any Chinese provocations. In February 2023, after a Chinese high-altitude balloon floated across U.S. airspace, the administration declassified details about it, in part to justify to the public its intense focus on competition with China and in part to signal to Beijing the U.S. intelligence community’s impressive technical capabilities. The Pentagon released a close-up photo of the balloon taken by a U-2 pilot, and officials explained to reporters that the U.S. government could track the object and had determined that it was loitering above sensitive military sites.

Later that month, the administration sought to warn Beijing that it was monitoring possible Chinese support for Russia’s war in Ukraine. CIA Director William Burns, undoubtedly with the approval of other senior officials, revealed in a televised interview that Beijing was considering offering Moscow lethal aid, adding, “We don’t see evidence of actual shipments of lethal equipment.” Burns clearly wanted to brush back the Chinese before they crossed a line they couldn’t uncross.

In the first few months of the war in Gaza, the Biden administration used intelligence disclosures to give Israel breathing space from mounting pressure about the destructiveness of its military campaign. Shortly after Hamas’s October 7, 2023, attack, Adrienne Watson, a spokesperson for the National Security Council, countered accusations that an Israeli bomb had struck a hospital in Gaza City, announcing that “overhead imagery, intercepts, and open-source information” suggested that the real culprit was an errant rocket fired by a terrorist group in Gaza. In November, the White House again came to Israel’s defense, with Kirby sharing a declassified intelligence assessment saying that Hamas was using hospitals as command-and-control nodes, weapons depots, and hideaways for Israeli hostages.

Strategic disclosures are set to become a durable feature of the U.S. foreign policy landscape. The Biden administration’s strategic downgrades have created an expectation on the part of the public, the media, and allies that there will be more to come, and it is unlikely that Biden or any of his successors will abandon the tool. The genie is out of the bottle.

THE COSTS OF CANDOR

But is any of this a good idea? Most policymakers seem to think so. For one thing, they have argued, disclosures have delivered results. Writing in these pages earlier this year, Burns argued that the Ukraine disclosures put Putin “in the uncomfortable and unaccustomed position of being on the back foot” and “bolstered both Ukraine and the coalition supporting it.” And it is reasonable to conclude that administration officials give at least partial credit to Burns’s disclosure about Beijing’s consideration of lethal aid to Russia for convincing Chinese leader Xi Jinping to not cross the line.

The second argument made by proponents of disclosures is that any transparency on the part of the secret state is good. Although Director of National Intelligence Avril Haines’s 2023 National Intelligence Strategy says nothing about strategic downgrades—a missed opportunity—it does endorse increased openness. Haines has elaborated on the idea, saying, “With the increasing importance of national security in our everyday lives, the more we can help to inform the public debate around such issues, the better.” This push for transparency is driven in part by the deluge of publicly available open-source intelligence, which invites doubts about the value of the intelligence community in a world where Bellingcat and other investigative groups seem to know as much as governments. (By the same token, however, open-source intelligence frees agencies to say more about what they know while jeopardizing less.) The push is also motivated by the public’s growing desire for government accountability in the wake of the intelligence failures behind the U.S. invasion of Iraq, revelations about the NSA’s industrial-scale information-collection capabilities, and Trump’s claims that the intelligence community and the rest of the “deep state” undermined his presidency. Disclosures, the argument runs, can rebuild public trust in the U.S. intelligence community, in part by demonstrating its value.

The most common criticism of disclosures is that they jeopardize intelligence sources and methods. If officials in a targeted regime know what the U.S. government knows about them, they can sometimes work backward to discover the source of that information—whether it be a tapped phone line, a cyber-exploit, or a member of the inner circle. They might shut down that channel, feed disinformation into it, or, in the case of human intelligence, arrest or harm the source. Some disclosures have undoubtedly led to a subsequent loss of intelligence. The Kennedy administration’s sharing of U-2 photographs of Cuba accomplished its intended statecraft goal but also revealed to the world for the first time just how sophisticated U.S. aerial surveillance was. Afterward, U.S. adversaries learned to better camouflage sensitive sites and improved their high-altitude air defense systems.

Authorized disclosures could result in more unauthorized ones.

But the intelligence community is acutely aware of these risks and works to mitigate them. Administrations have tended to declassify broad analytic judgments that carry little risk to sources and methods, leaving out the sensitive intelligence nuggets that could allow the source to be identified. The intelligence community, for its part, is not shy about standing firm and refusing a policy request for a particular disclosure when the risks are just too high. One of us, Morell, was involved in declassifying information for Powell’s presentation to the UN Security Council; concerned about sources and methods, the CIA denied some of Powell’s (and the White House’s) requests for declassification.

Less well understood are the subtler risks to sources and methods. One is that human sources will get skittish about divulging information, no matter what their handlers promise in terms of security or reward. Many CIA case officers—including one of us, Gioe—have had the experience of listening to their assets express grave concern about the growing volume of intelligence that has gone public, whether through an illegal mass leak or an authorized disclosure, and ask if the information they provide might go public, too. Some assets have even walked away in the aftermath of prominent leaks or disclosures. And it is impossible to calculate how many would-be assets have changed their mind as a result of them.

Another risk to sources and methods is that authorized disclosures could result in more unauthorized ones by raising questions about just how appropriate it was to classify something in the first place. It can be perfectly reasonable to conclude that the national security benefits of a given disclosure outweigh the risks. Still, the nuances of such judgment calls could be glossed over by leakers, who may see coordinated and authorized disclosures as justification or cover for their own reckless revelations. Snowden, for instance, complained in his autobiography, “It is rare for even a day to go by in which some ‘unnamed’ or ‘anonymous’ senior government official does not leak, by way of a hint or tip to a journalist, some classified item that advances their own agenda or the efforts of their agency or party.” In other words, if government officials can release intelligence when it suits them, why can’t anyone else?

A separate risk is that some of the information released turns out to be wrong, damaging the reputation of the intelligence community. Although intelligence agencies were right about Russia’s intention to invade Ukraine—even getting the timing right—such high accuracy is not the norm. (Indeed, they were wrong to predict that the Ukrainians wouldn’t last long in battle, a judgment that the White House almost certainly never wished would go public but ended up leaking anyway.) Despite an annual budget of around $100 billion, the U.S. intelligence community does not have a crystal ball and cannot supply evidence fit for a courtroom.

For one thing, intelligence on almost any issue is by nature imperfect and fragmentary; adversaries go to great lengths to protect the information the United States is after and, in some cases, are actively deceiving Washington. For another thing, intelligence is dynamic. During the 2021 U.S. withdrawal from Afghanistan, for instance, the intelligence community was continually revising its assessments of the Afghan government’s ability to resist the Taliban. Critics claimed that the chaotic exit was partly the result of intelligence agencies’ failure to predict the Taliban’s swift victory. That may be true, but the situation was changing by the hour, and it is inherently hard to predict if or when an unstable system will collapse. Intelligence failures happen for any number of reasons, legitimate and otherwise, but when they do occur, the reputation of the intelligence community gets dented. In a world of routine strategic downgrades, it should expect some dents.

The greatest risk with disclosures is the politicization of intelligence. In its most benign form, politicization takes the form of releasing accurate but incomplete information. Since the point of disclosures is to advance an administration’s policy, it is only natural that officials will select the intelligence that does so and keep classified any intelligence to the contrary. (When the Biden administration released an estimate of Russian casualties in Ukraine, it notably remained mum on the Ukrainians’ own high losses.) This preference is acceptable when trying to influence a foreign adversary, but not when the audience is the American people. Informing citizens is a laudable, apolitical act; trying to shape their views by cherry-picking intelligence is not. In the lead-up to the Iraq war, Tenet fielded competing requests from members of Congress to declassify only those portions of intelligence assessments that buttressed a particular argument. One camp, for example, wanted to release a judgment that Saddam was unlikely to initiate a terrorist attack against the United States, whereas another wanted to release one that Saddam was likely to use weapons of mass destruction if he felt cornered. Tenet did the right thing by declassifying both judgments.

In the more egregious form of politicization, policymakers actively misrepresent the intelligence they disclose or stake a position beyond what it can support. This has happened too frequently to dismiss it as a minimal risk. In 1964, for example, President Lyndon Johnson used a confrontation between U.S. and North Vietnamese naval forces in the Gulf of Tonkin to push Congress to grant him more power in prosecuting the Vietnam War. Although there had been just one incident, in a speech to the American people, Johnson claimed that there were two—deliberately going well beyond what ambiguous intelligence reports had suggested. In the Bush administration, several senior officials, intent on making a stronger case for invading Iraq, publicly stated that Saddam’s regime had an ongoing relationship with al Qaeda—the exact opposite of what the CIA had concluded.

Amid these cautionary tales, one historical example offers a model for disclosure: the Bush administration’s 2008 revelations about a Syrian nuclear reactor, apparently built with North Korean help, that an Israeli airstrike had leveled a year before. The disclosure was intended to strengthen efforts to persuade the North Koreans to provide a full accounting of their nuclear weapons activity and efforts to end Iran’s uranium enrichment activities. In a declassified briefing to reporters, CIA Director Michael Hayden outlined the intelligence surrounding the discovery, making it clear what the intelligence community knew and didn’t know, as well as how confident it was about its judgments. He said that analysts had “high confidence” that the building destroyed by Israel was indeed a nuclear reactor, “medium confidence” that North Koreans had assisted in building it, and only “low confidence” that it was part of a Syrian nuclear weapons program. The last caveat was the kind that policymakers typically want to strip out, but Hayden wisely put it in. His specificity in connecting each judgment to a corresponding confidence level made it harder for anyone to politicize the information.

USE WITH CAUTION

The risks from strategic downgrades are real and, given their accelerating use, growing. The decision to disclose intelligence is a policy call, and in making it, officials have to strike a delicate balance, supporting a given policy goal while protecting sources and methods and maintaining analytical integrity. As Jon Finer, Biden’s deputy national security adviser, has observed, strategic downgrades “must be wielded carefully within strict parameters and oversight.” So what should those guardrails be?

First, any disclosure should pose little threat to intelligence sources or methods—a finding that must reflect the consensus of the intelligence community. The decision to disclose should be made by the director of national intelligence and only after a full consideration of the risks to sources and methods. Disclosures that do reveal sources are usually a judgment call, but a tie shouldn’t go to the policy runner. One rule of thumb is to release analytical judgments but not the underlying raw intelligence on which they are based. This is what the Obama administration did with the intelligence community’s report about Russian meddling in the 2016 presidential election. This approach represents a sort of halfway house for disclosure, but of course it will not satisfy those skeptics who understandably wish to see the underlying evidence before believing the intelligence community’s conclusions. Nevertheless, for many, even all the information would not be enough, and in any case, the imperative must be to protect sources and methods.

Second, judgments released publicly should have a high likelihood of being correct. If they turn out to be wrong, the intelligence community’s reputation will suffer and the effectiveness of future disclosures will be undermined, since there would be a historical basis to doubt them. One remedy would be to release only high-confidence judgments. In 2023, Burns signaled that he had done just that when he said he was “confident” that Chinese leaders were considering providing Russia with lethal aid, a word choice that suggested analysts believed there was a high likelihood their judgment was correct. Another option would be to follow Hayden’s example and disclose intelligence of various levels of confidence but make clear which conclusions enjoy which level.

Disclosures have a mixed record and are probably less successful than officials believe.

Third, in a world of disinformation and spin, the release of intelligence must represent the truth or, more precisely, what the intelligence community assesses to be true. Although it may be tempting to embed disinformation in a disclosure, that line should never be crossed. Nor should officials attempt to spin the intelligence in any way to create a misleading impression. And crucial caveats should always be included, since withholding them creates the illusion of certainty.

Fourth, a disclosure should have to pass a common-sense test: that there be a reasonable chance it will have the intended effect. Disclosures have a mixed record and are probably less successful than officials believe. One requirement for success is that a strategic downgrade be connected to an overarching strategy involving the rest of the U.S. government; if it isn’t, its chances of working are markedly reduced. The Biden administration’s disclosures about the impending invasion of Ukraine, for instance, had some positive effect, but they could not compensate for years of poor policies, such as the failure to impose tough sanctions or give Kyiv enough military aid after Putin seized Crimea.

Although bureaucratic rules are rarely the best solution to real-world problems, there is one that would help here. The director of national intelligence should issue an intelligence community directive (the intelligence community’s equivalent of an executive order) stipulating that a disclosure can be made only after she has signed a memorandum that addresses all the guardrails. This requirement would not only instill discipline but also create a record of important decisions. The office of the director of national intelligence could then develop an internal dataset—trackable over time and available to her successors—to assess the short- and long-term effects of disclosures.

BRAVE NEW WORLD

The conundrum of strategic downgrades is but one of many challenges facing the U.S. intelligence community. The list is long: how to recruit spies in a world of ubiquitous technical surveillance, how to collect signals intelligence in a world of decentralized telecommunications and computing, how to sift through mountains of data in a world of open-source information, and how to hire and retain the best and the brightest in a world of declining trust in government. And all these difficulties are set against the backdrop of great-power competition, with China, Russia, and other authoritarian countries working every day to threaten the United States’ democracy, prosperity, and security.

Officials outside intelligence agencies, for their part, generally do not approach disclosures with the same caution as the people serving inside them. Policymakers’ natural confidence and enviable optimism about the efficacy of their own actions may invite them to focus on the upsides of disclosures while ignoring or rationalizing away the dangers. Their desire to maximize the policy utility of secret intelligence may lead them to resist efforts to add new restraints to the disclosure process.

Given all these pressures, it would be tempting for policymakers and intelligence practitioners alike to throw up their hands and decide to manage strategic declassification in an ad hoc way. But that would be a mistake. The point of no return has been passed, and intelligence is being released faster than norms can be created. If the process for disclosures is not handled with utmost care, the United States could diminish the unparalleled advantage in statecraft and national security it derives from a crucial pillar of American power: the U.S. intelligence community.