The formerly classified fact that one metric ton of plutonium metal was to be moved from the Savannah River Site in 2019 for use in nuclear weapon pit production at Los Alamos was declassified in 2018. This recently disclosed declassification decision was one of a handful of such actions that are taken each year by the Department of Energy.
DOE is required to perform “continuous review” of information that is classified under the Atomic Energy Act (sec. 2162) and to periodically determine which information can be removed from the category of Restricted Data and declassified. And so it does, every now and then.
Some of the resulting declassification decisions pertain to specific events, like the transfer of nuclear material from one site to another. Others are narrowly technical, like the declassification of “the static and dynamic equations of state for 71 < Z < 90 for pressures > 10 Mbar.” (Z is the atomic number, where 71 is Lutetium and 90 is Thorium.)
In one anomalous case, the scope of the declassification action itself was redacted and remains undisclosed. This is somewhat hard to understand but DOE apparently holds that, having been declassified, the entire subject of this action now falls within the category of “unclassified controlled nuclear information” which is exempt from disclosure.
Declassification decisions under the Atomic Energy Act through last year were released under the Freedom of Information Act.
* * *
The Federation of American Scientists last week renewed its petition to the Department of Energy and the Department of Defense to declassify the current number of weapons in the U.S. nuclear stockpile and the number that have been dismantled.
“US nuclear weapons policy should be conducted on the basis of accurate public information to the extent possible,” the FAS petition said. “Declassification of stockpile data supports a factual deliberative process in Congress and elsewhere.”
“We will begin the process of evaluating your proposal and conducting the necessary coordination,” replied Nick Prospero, the acting director of the Office of Classification at the Department of Energy.
Update: On October 5, the Department of State and the National Nuclear Security Administration released annual stockpile and dismantlement figures through FY 2020.
* * *
Another FAS petition, filed in 2018, to declassify the size of the current US inventory of highly enriched uranium has lately received a favorable response from the Department of Energy.
“The program office has indicated they are ready to support the declassification request,” said Andrew Weston-Dawkes, then-director of the DOE Office of Classification, on September 8. “I suspect there is a good amount of work to collect and process the HEU data so hopefully we can provide an update on status in a couple of months.”
Legislative measures to improve the process of declassifying classified national security information were introduced by Sen. Ron Wyden in the pending intelligence authorization act for FY2022. But they were included in the classified annex so their substance and import are not publicly known.
“I remain deeply concerned about the failures of the Federal Government’s obsolete declassification system,” Sen. Wyden wrote in a statement that was included in the new Senate Intelligence Committee report on the intelligence bill. “I am therefore pleased that the classified annex to the bill includes several amendments I offered to advance efforts to accelerate declassification and promote declassification reform.”
But the nature of those amendments has not been disclosed. “As absurd as it is to be opaque about the topic of declassification I’m afraid I can’t tell you more right now,” a Committee staffer said. “Sorry.”
“Putting aside the irony of declassification amendments found only in a classified annex, I can confirm that we’re tracking it,” said an intelligence community official.
It requires some effort to think of declassification reforms that could themselves be properly classified. The idea seems counterintuitive. But there are agency declassification guides that are classified because they detail the precise boundaries of classified information. And any directive to declassify an entirely classified topical area would have to begin by identifying the classified subject matter that is to be declassified.
“True perfection seems imperfect,” says the Tao Te Ching (trans. Stephen Mitchell), and “true straightness seems crooked.” Still, some things are truly crooked.
For the past three years, the Trump Administration refused to provide an annual tally of the number of nuclear weapons in the U.S. nuclear stockpile or the number of weapons that had been dismantled each year, though that had been the practice under the Obama Administration and through 2017.
The Federation of American Scientists asked the Biden Administration to restore the prior level of disclosure and to report the missing stockpile and dismantlement numbers for 2018, 2019 and 2020.
In a petition submitted to the Department of Energy, FAS noted that President Biden had directed agencies in a February 4 directive to adopt “the highest standards of transparency.”
“By terminating the annual disclosure of stockpile information, the prior Administration retreated from ‘the highest standards of transparency’ that previously prevailed,” the FAS petition said.
The Department of Energy acknowledged receipt of the request. Declassification and disclosure of the requested stockpile data will require concurrence of both DOE and DoD. Historically, DOE has long been willing to disclose such information, while DoD has often resisted release.
* * *
According to FAS estimates by Hans M. Kristensen and Matt Korda, the current U.S. nuclear stockpile is around 3,800 weapons, and the number of weapons dismantled is probably around 300-350 per year. For more detail, see the FAS Nuclear Notebook on United States nuclear weapons, 2021, published by the Bulletin of the Atomic Scientists.
Although the unofficial estimates are probably quite accurate, formal declassification remains desirable both in order to curtail improper secrecy and to enable government officials to freely address the subject in open public forums.
Update (October 5, 2021): The Department of State and the National Nuclear Security Administration have now released annual stockpile and dismantlement figures through FY 2020.
The declassification process has been overwhelmed by the flood of classified records awaiting review, said Sen. Ron Wyden last week. “I intend to push the Director of National Intelligence to fix a broken declassification system,” he said.
One highly effective way to begin fixing the declassification system would be to set a maximum period of time that information can remain classified. This maximum lifetime for classification is sometimes referred to as a “drop dead” date.
Records that reach the drop dead date would not require review and they would not even need to be declassified in any formal way. Their prior classification status would simply lapse without any further processing.
This approach was favored by the Information Security Oversight Office (ISOO) during the Clinton Administration.
“We’re looking at the idea of a ‘drop dead’ date — [meaning that] anything that is x number of years old is declassified,” the late Steven Garfinkel said in a 1992 interview, when he was ISOO director. But “x” is “going to be a lot longer than twenty years,” he said then.
How long should it be? Mr. Garfinkel thought that a forty year duration for classification would be the optimal period for maximizing declassification while minimizing risk.
“When we looked at material that was 40 years old or older, we were ending up declassifying just about every bit of it, far more than 99 percent of it,” Mr. Garfinkel told a Defense Department advisory panel in 1996. In fact, he recalled, the first draft of the Clinton executive order actually “had a 40-year drop dead date.”
But under pressure from openness advocates (including myself), who argued that 40 years was too long to wait, the final Clinton executive order instead adopted a 25 year period for “automatic declassification.” Unfortunately, however, in doing so it also provided nine exemptions from declassification. This negated the concept of a drop dead date. It entailed painstaking review by agencies to locate and redact exempted information. The resulting declassification process was anything but “automatic” — as Steve Garfinkel had anticipated.
But the idea of a drop dead date is still a good one that has many positive features.
It would put a temporal boundary on the classification system. It would complement current efforts to prioritize the declassification of documents that are in high public demand by ensuring that all other documents will also ultimately be declassified. It would terminate the spiral of repeated reviews of the same partially redacted documents. And it would relieve agencies of a massive bureaucratic burden — at literally no cost — that will be difficult or impossible to overcome in any other practical way.
Arguably, a drop dead date is already implicit in current policy. Executive Order 13526 (sect. 1.5d) states that “No information may remain classified indefinitely.” All that is needed is to act on that established principle and to set a definite limit on the duration of classification.
A drop dead date for classification would not solve all declassification problems, even for the oldest classified documents. Information concerning nuclear weapons that is still classified under the Atomic Energy Act cannot be declassified by fiat, no matter how old it is. Certain classified information that is controlled by international agreement or treaty would also have to be protected regardless of its age. But these represent a small fraction of the massive accumulation of old records awaiting declassification.
It is possible to imagine a document that is 40 or more years old that nevertheless poses some kind of articulable national security threat. If this proved to be a real concern, then some allowance for the rare exception to the drop dead date could be made. But considering that current US government information systems have been widely penetrated by foreign actors, any residual threat from applying a drop dead date to half-century old documents could reasonably be deemed tolerable — especially if it helped agencies to finally overcome their declassification backlogs.
“I agree that current policies, practices, and technologies cannot keep pace with the amount of classified information generated each year,” said Avril D. Haines, the new Director of National Intelligence, last week in response to pre-confirmation questions from the Senate Intelligence Committee.
“I look forward to examining these challenges and identifying what changes may be warranted,” she said.
Classified records that turn 25 years old this year will be automatically declassified on December 31 — despite requests from agencies to extend the deadline due to the pandemic — unless the records are reviewed and specifically found to be subject to an authorized exemption.
Mark A. Bradley, the director of the Information Security Oversight Office, notified executive branch agencies last week that there is no basis in law or policy for deferring the automatic declassification deadline.
“Several agencies have expressed concerns that, due to diminished operational capacity and capability, they would likely be unable to complete declassification reviews of their 25-year old classified permanent records before the onset of automatic declassification on December 31, 2020. These agencies have requested some form of relief, such as a declassification delay or waiver,” Mr. Bradley said in his November 20 letter.
But the executive order that governs declassification and the implementing regulations “do not permit the declassification delays or waivers requested in this instance,” he wrote.
Mr. Bradley advised agencies “to adopt a risk-based approach and prioritize the review of their most sensitive records” in order to identify the most important information that might be exempt from automatic declassification.
But the fact remains that any “Originating agency information in 25-year old permanent records that are not reviewed prior to December 31, 2020 will be automatically declassified,” he wrote.
Mr. Bradley’s letter emphasized that automatic declassification applies only to information in records held by the originating agency, but not to information that originated with other agencies. Such other agency “equity” information is supposed to be referred to those agencies for their subsequent review.
Yet although the letter does not mention it, under the terms of the executive order (sec. 3.3d(3)) the identification of information generated by another agency is also supposed to be completed in advance of the December 31 deadline. It is unclear whether an agency’s failure to identify information for referral to other agencies prior to the deadline would nullify the referral and eliminate the opportunity for subsequent review.
While records that have been automatically declassified can in principle be reclassified, that is easier said than done. Though automatic declassification can be performed in bulk, reclassification is only permitted on a document by document basis, requiring in each case a written justification by the agency head.
Last week, the State Department announced the publication of the latest volume of the Foreign Relations of the United States series documenting the Iran hostage crisis of 1979-80.
Few if any of the newly published records were subject to automatic declassification. Instead, “The declassification review of this volume . . . began in 2010 and was completed in 2018,” the editors wrote.
On November 8 Donald J. Trump Jr., the President’s oldest son, tweeted: “DECLASSIFY EVERYTHING!!!” adding “We can’t let the bad actors get away with it.”
This was not an actual policy proposal and it was not seriously intended for classification officials or even for Trump’s own father, who as President is the one ultimately responsible for classification policy.
Rather, it was directed at Trump Jr.’s 6.4 million Twitter followers, telling them that classification is a corrupt process that protects “bad actors” and that must therefore be discredited and dismantled. It’s a juvenile notion but not, given the size and malleability of Trump’s audience, an inconsequential one.
To the extent that national security classification is in fact required, for example, to protect advanced military technologies, the conduct of diplomacy or the collection of intelligence, it is important to establish and maintain the legitimacy of classification policy. For the same reason, abuse of classification authority can itself be a threat to national security.
The current executive order on classification policy (sect. 1.7a(1)) directs that “in no case shall information be classified . . . in order to conceal violations of law.”
But this is merely a limitation on the classifier’s mental state — which is unverifiable — and not on classification itself. It is entirely permissible for classified information to conceal violations of law, according to a judicial interpretation of the executive order, as long as the information is not classified with that specific purpose (“in order to”) in mind. This is a standard that has never been enforced and that is probably unenforceable.
So one step that the incoming Biden Administration could take to enhance the integrity and accountability of classification policy would be to direct that classification may not conceal violations of US law at all, whether or not that is the intent of classifying. (It is probably necessary to specify “US” law since classified intelligence collection may often involve the violation of foreign laws.)
Donald Trump is the first president since George H.W. Bush who made no formal changes to the executive order on classification policy.
Instead, Trump often defied or disregarded existing classification and declassification policies, withholding previously public information (e.g. the number of nuclear warheads dismantled each year) and disclosing normally classified information (e.g. an actual application for counterintelligence surveillance) when it advanced his political interests to do so.
But it seems that arbitrary secrecy combined with selective declassification is not the way to stop “bad actors.”
Last January the Trump Administration formally notified Congress under the War Powers Act of a US drone strike that killed Iranian Maj. Gen. Qasem Soleimani.
But unlike all known prior War Powers Act notifications, the report on the Soleimani killing was classified in its entirety. (Previous reports sometimes included a classified annex together with the unclassified notification.)
Senator Chris Murphy (D-Conn.) said that was unacceptable. “There’s a veil being pulled over the foreign policy of this country,” he told the Washington Post. See “Six months later, Democrats keep working to unearth a big national security secret” by Greg Sargent, The Washington Post Plum Line, July 21, 2020.
Senator Murphy asked the White House to reconsider the classification. “It is critical that decisions regarding the use of force consistent with the War Powers Act be provided in unclassified form to the American people,” he wrote. He received no response.
So he turned to the Interagency Security Classification Appeals Panel (ISCAP), a group of executive branch agency representatives that is authorized by executive order to decide appeals of challenges to classification.
The initiative failed. Last month the ISCAP said that it would not consider such an appeal from Senator Murphy or from any other member of Congress.
The ISCAP refusal leaves the War Powers Act report on Soleimani fully classified and it keeps the public in the dark about the asserted legal and factual basis for killing him. But it highlights an important gap in classification policy that could be corrected in a new Administration and a new Congress.
* * *
When information is classified improperly or unnecessarily, the opportunities for correcting such actions are quite limited.
A provision for government employees to formally challenge the classification of certain information was introduced in President Clinton’s 1995 executive order 12958 (section 1.9) and has remained in effect until the present (executive order 13526, section 1.8). The provision states:
“Authorized holders of information who, in good faith, believe that its classification status is improper are encouraged and expected to challenge the classification status of the information. . . .”
Importantly, this provision was not intended as a courtesy or a privilege. In fact, it was not intended for the sake of the challengers at all. Rather, the purpose of such classification challenges was to promote the integrity of the classification system and to help make the system self-correcting, as far as possible. That’s why potential challengers are “encouraged and expected” to present challenges even if they don’t personally care about the issue at all.
There were 954 such challenges in fiscal year 2016, according to the Information Security Oversight Office, and 167 of those resulted in the classification being overturned in whole or in part. In FY 2017, there were 721 challenges, 58 of which led to changes in classification.
No member of Congress had ever invoked this provision before. But Senator Murphy had some reason to believe that such a classification challenge could be effective in the case of the Soleimani war powers report.
The sticking point was the definition of “authorized holders of [classified] information,” who are the only ones that can present a classification challenge under the executive order.
One would suppose that a member of Congress who is in possession of a classified report that was officially provided to him or her by the executive branch would certainly qualify as an “authorized holder.” In fact, the executive branch has a binding legal obligation to provide certain classified defense and intelligence information to Congress.
But it turns out that the executive order (in section 6.1c) narrowly defines an “authorized holder of classified information” as one who has been vetted by an agency and found eligible for access. (Oddly, this limiting definition was only added in 2009.) Since Members of Congress are cleared for classified information by virtue of their office and do not undergo agency vetting, they are not “authorized persons” for purposes of the executive order.
This does not make any sense from a policy point of view. Just as executive branch employees and contractors are “encouraged and expected” to point out potential errors in classification, so should Members of Congress be, and for the same reasons.
But the classification challenge procedure is constrained by the language of the executive order, said Mark Bradley, director of the Information Security Oversight Office and executive secretary of the ISCAP.
“We have to do what the Order says, not what we want,” said Mr. Bradley, who early in his career served as an aide to Senator Daniel P. Moynihan.
* * *
Mr. Bradley suggested that Senator Murphy could direct his challenge to the Public Interest Declassification Board, which unlike the ISCAP is specifically authorized to review congressional challenges to the classification of certain records.
But the PIDB is a much weaker body than the ISCAP. While the ISCAP can “decide” on classification challenges (subject to appeal), the PIDB can only review and “recommend.” And while the ISCAP has actually overturned existing classifications on numerous occasions, no PIDB recommendation has ever had the same effect.
The PIDB did previously handle one congressional request for declassification review, said John Powers of the ISOO and PIDB staff, in or around 2014. For the most part, the subject document in that case turned out to be properly classified, substantively and procedurally, in the PIDB’s view. But the Board forwarded a limited redaction proposal that would have allowed partial release to the Obama White House for consideration. The White House did not act on it.
Senator Murphy turned to the PIDB to request declassification review of classified intelligence concerning foreign interference in the upcoming US elections, the Washington Post reported yesterday.
* * *
The statement by ISOO director Mark Bradley cited above — “We have to do what the Order says, not what we want” — is worth further consideration.
What he was saying is that those who are responsible for enforcing checks and balances have to follow a code of conduct and have to adhere to a set of principles, whether or not they personally agree with the outcome in a particular case.
The problem is that those who abuse the system to classify (or sometimes to selectively declassify) information improperly recognize no such constraint. This discrepancy is vexatious.
It means that the checks and balances of the current system are most effective when they are least necessary. When everyone is acting in good faith and with an honest commitment to shared (constitutional) values, most disagreements can be resolved over time. Some compromise is usually possible.
But when good faith and principled self-restraint are lacking, and one side aims to maximize its power at any cost, the current structure of checks and balances has proved to be largely helpless.
Even if the ISCAP had agreed to consider Senator Murphy’s classification challenge, and if it had actually agreed with him that all or part of the War Powers Act notification concerning the Soleimani killing was not properly classified, that might not have been the end of the story.
“Panel decisions are committed to the discretion of the Panel,” according to the executive order (sect. 5.3e), “unless changed by the President.” But that means that a hypothetical ISCAP decision to declassify the notification could be overruled by the same White House that classified the whole thing in the first place.
So while good policies are necessary, they are not enough. For our constitutional system of government to work, we also need officials who are, if not the “angels” that James Madison spoke of, at least dedicated public servants who share a common purpose.
Senator Murphy’s office said that he would soon introduce legislation to authorize and require the ISCAP to consider classification challenges from Congress.
* * *
The current infrastructure for declassifying classified records that are no longer sensitive is already being overwhelmed by a deluge of historical records that are accumulating faster than they can be processed. This situation was discussed in a September 9 hearing before the Senate Select Committee on Intelligence and is the subject of new legislation (S. 3733) introduced by Senators Wyden and Moran.
That is an issue of efficiency and productivity that probably has a technological solution, as the Public Interest Declassification Board has argued.
A harder problem is over-classification, in which information is classified improperly or unnecessarily, or at a higher level than is warranted. Such classification errors can be corrected, at least hypothetically, through classification challenges, Freedom of Information Act requests, and other means.
A still harder problem concerns information that is properly classified — in the sense that it meets the criteria of the executive order — but nevertheless belongs in the public domain because of its fundamental policy importance. Examples include classified reports of torture, mass surveillance, or foreign election interference.
To the extent that such information is “properly classified” in a formal sense, it is currently beyond the reach of the Freedom of Information Act, mandatory declassification review, or classification challenges. When it does become public, that is often due to unauthorized disclosures. While agency heads may declassify classified information in the public interest as a matter of discretion (under section 3.1d of the executive order), they rarely do so and there is no mechanism for asking or inducing them to.
So along with adequate basic functionality and improved procedures for challenging improper classification, any future classification system also needs to tackle the problem of “properly classified” information that should not be classified.
Obsolete secrecy procedures and growing political abuse have left the national security classification system in a state of disarray and dysfunction.
Most government agencies “still rely on antiquated information security management practices,” according to a new annual report from the Information Security Oversight Office (ISOO). “These practices have not kept pace with the volume of digital data that agencies create.”
“Agencies are not applying or testing advanced technologies that would enable more precise classification and declassification, facilitate information sharing, and improve national security,” the ISOO report to the President said. “Classification and declassification actions are still performed manually, which is neither sustainable nor desirable in the digital age.”
“As the volume of records requiring [declassification] review increases, agencies are making more errors, putting Classified National Security Information at risk and eroding trust in the system,” ISOO said.
As damning as these and other ISOO findings may be, they hardly begin to capture the crisis of credibility that is facing the classification system today.
An effective classification system depends on a presumption of good faith on the part of classifiers, checked by independent oversight, and some consensual understanding of the meaning of national security. All of these factors are in doubt, absent, or undergoing swift transformation. Meanwhile, classification today is openly wielded as an instrument of political power.
“Conversations with me, they’re highly classified,” said President Trump last week. “I told that to the Attorney General before. I will consider every conversation with me, as President, highly classified.”
That remark is a wild departure from previous policy. However broadly it may have been construed in the past, classification was always supposed to apply to information that was plausibly related to national security (a necessary condition, though not a sufficient one). Even the most sensitive conversations with the President about tax policy or health care, for example, could not have been considered classified information.
In this case, President Trump was objecting to the publication of the new book by former national security adviser John Bolton, which he dismissed at the same time as a “compilation of lies and made-up stories.”
But Bolton’s lies, if that’s what they were, would not normally qualify as classified information either.
In principle, it’s possible that “lies and made-up stories” could be classified, though only to the extent that they were generated by the government itself (perhaps in the form of cover stories, or other official statements of misdirection). But any lies that Bolton might tell on his own are beyond the scope of classification, since they are not “owned by, produced by or for, or. . . under the control of” the US Government, as required by the executive order on classification.
President Trump may or may not understand such rudiments of national security classification. But by twisting classification policy into a weapon for political vendettas, the President is discrediting the classification system and accelerating its disintegration.
As for Bolton, the astonishing fact is that he is the second of Trump’s national security advisors (after Gen. Michael Flynn) to be accused of lying and criminal activity.
“If the [Bolton] book gets out, he’s broken the law,” the President said. “And I would think that he would have criminal problems.” Indeed, a court said on Saturday that Bolton might have “expose[d] himself to criminal liability.”
Second only to the President, the national security advisor is really the principal author and executor of classification policy. So when NSAs like Flynn and Bolton are disgraced (or worse), their disrepute reflects upon and attaches to the classification system to some degree.
Ironically, Mr. Bolton was more attentive to and more engaged in classification policy than many of his predecessors. He makes a tacit appearance (unnamed) in the new ISOO annual report, which notes: “In FY 2018, the ISCAP [Interagency Security Classification Appeals Panel] received a request from the National Security Advisor to resolve a declassification dispute between the Departments of Defense and State.” That action, to Bolton’s credit, freed up all or parts of 60 documents for publication in the Foreign Relations of the United States series, over the objections of the Department of Defense.
The way to begin restoring credibility to classification policy is not hard to envision, though it may be difficult or impossible to implement under current circumstances. Like law enforcement, the classification system needs to be insulated from partisan political interference. Classification policy needs to adhere to well-defined national security principles (though the scope and application of these principles will be debatable). A properly functioning classification and declassification system will prove its integrity by sometimes producing outcomes that are politically unwelcome or inconvenient to the Administration. And since errors are inevitable, the classification system also requires a robust oversight and error-correction process.
* * *
Last week, the Senate Intelligence Committee blocked an effort by Senator Ron Wyden to restructure and strengthen the declassification system. A Wyden amendment to the FY 2021 intelligence authorization act would have designated the Director of National Intelligence as the Executive Agent for declassification, tasking him to establish and carry out government-wide declassification requirements. The Wyden amendment failed 7-8 with all Republican members opposed.
By rejecting his amendment (without offering any alternative), the Committee “failed to reform a broken, costly declassification system,” Sen. Wyden said in a dissenting statement appended to the June 17 report on the intelligence bill.
* * *
While dismissing concerns about classification policy, the Senate Intelligence Committee roused itself to address the threat from unidentified flying objects, an issue that it said requires more focused government attention.
The Committee called on the Director of National Intelligence to provide detailed reporting on “unidentified aerial phenomena (also known as ‘anomalous aerial vehicles’), including observed airborne objects that have not been identified.”
“The Committee remains concerned that there is no unified, comprehensive process within the Federal Government for collecting and analyzing intelligence on unidentified aerial phenomena, despite the potential threat,” the new Committee report said.
How can the national security classification and declassification system be fixed?
That depends on how one defines the problem that needs fixing. To the authors of a new report from the Public Interest Declassification Board (PIDB), the outstanding problem is the difficulty of managing the expanding volume of classified information and declassifying a growing backlog of records.
“There is widespread, bipartisan recognition that the Government classifies too much information and keeps it classified for too long, all at an exorbitant and unacceptable cost to taxpayers,” said the PIDB, a presidential advisory board. Meanwhile, “Inadequate declassification contributes to an overall lack of transparency and diminished confidence in the entire security classification system.”
The solution to this problem is to employ technology to improve the efficiency of the classification and declassification processes, the PIDB said.
“The time is ripe for envisioning a new approach to classification and declassification, before the accelerating influx of classified electronic information across the Government becomes completely unmanageable,” the report said. “The Government needs a paradigm shift, one centered on the adoption of technologies and policies to support an enterprise-level, system-of-systems approach.”
See A Vision for the Digital Age: Modernization of the U.S. National Security Classification and Declassification System, Public Interest Declassification Board, May 2020.
The report’s diagnosis is not new and neither is its call for employing new technology to improve classification and declassification. The PIDB itself made similar recommendations in a 2007 report.
Recognizing the persistent lack of progress to date, the new report therefore calls for the appointment of an Executive Agent who would have the authority and responsibility for designing and implementing a newly transformed classification system. (The Director of National Intelligence, who is already Security Executive Agent for security clearance policy, would be a likely choice.)
Those who care enough about these issues to read the PIDB report will find lots of interesting commentary along with plenty to doubt or disagree with. For example, in my opinion:
* The useful idea of appointing an Executive Agent is diminished by making him or her part of an Executive Committee of agency leaders. The whole point of creating a “czar”-like Executive Agent is to reduce the friction of collective decision making and to break through the interagency impasse. An Executive Committee would make that more difficult.
* The PIDB report would oddly elevate the Archivist of the United States, who is not even an Original Classification Authority, into a central role “in modernizing the systems used across agencies for the management of classified records.” That doesn’t make much sense. (An official said the intended purpose here was merely to advance the mission of the Archives in preserving historical records.)
* The report equivocates on the pivotal question of whether or not (or for how long) agencies should retain “equity” in, or ownership of, the records they produce.
* The report does not address resource issues in a concrete way. How much money should be invested today to develop the recommended technologies in order to reap savings five and ten years from now? It doesn’t say. Who should supply the classified connectivity among classifying agencies that the report says is needed? Exactly which agency should request the required funding in next year’s budget request? That is not discussed, and so in all likelihood it is not going to happen.
But the hardest, most stubborn problem in classification policy has nothing to do with efficiency or productivity. What needs updating and correcting, rather, are the criteria for determining what is properly classified and what must be disclosed. And since there is disagreement inside and outside government about many specific classification actions — e.g., should the number of US troops in Afghanistan be revealed or not? — a new mechanism is needed to adjudicate such disputes. This fundamental issue is beyond the scope of the PIDB report.
The Public Interest Declassification Board will hold a virtual public meeting on June 5 at 11 am.
The Department of Defense must explain by early next year how it is going to meet its obligations to declassify a growing backlog of classified records, Congress said this week.
A provision (sect. 1759) in the new House-Senate conference version of the FY2020 national defense authorization act requires the Pentagon to prepare a report including:
* a plan to achieve legally mandated historical declassification requirements and reduce backlogs;
* a plan to incorporate new technologies, such as artificial intelligence, that would increase productivity and reduce the cost of implementing such a plan;
* a detailed assessment of the declassified documents released in the past three years along with an estimate of how many will be released in the next three years;
* other policy and resource options for reducing backlogs of classified documents awaiting declassification.
While the new legislative language is a welcome acknowledgment of a persistent problem, it does not by itself significantly advance a solution. In particular, the legislation does not authorize any new funds for declassification or for development of new declassification technologies, which are not yet mature. Nor does it define an alternative in the event that DoD proves unable to meet its declassification obligations.
In a prior draft adopted by the House of Representatives, the CIA and the State Department would also have been required to prepare similar reports. But those requirements were dropped in the final bill.
“The U.S. government’s system for declassifying and processing historical records has reached a state of crisis,” wrote William Burr of the National Security Archive lately. See “Trapped in the Archives,” Foreign Affairs, November 29, 2019.
Update (October 2021): The required report on Reducing the Backlog in Legally Required Historical Declassification Obligations of the Department of Defense was released in October 2021.
A bill introduced by Senator Ron Wyden would require the FBI and the Director of National Intelligence to declassify “any and all information” regarding actions by the government of Saudi Arabia to assist Saudi nationals who are accused of crimes in the United States to flee the country.
As Senator Wyden explained last week, the bill was prompted by multiple cases — 19 over the past 7 years, according to The Oregonian newspaper — in which Saudi nationals who were awaiting trial on charges such as manslaughter and sexual assault were able to flee the U.S. and escape trial.
The bill is categorical in its requirements. It does not “ask” for declassification. It does not call for “declassification review” or allow for any exercise of discretion. It does not mention any possible exceptions to declassification for intelligence sources and methods, foreign government information, or other national security-related material. None of that.
Instead, it quite simply states: the FBI and the DNI “shall declassify any and all information” related to the possible role of the Saudi government in this matter.
Today’s national security classification system “relies on antiquated policies from another era that undercut its effectiveness today,” the Information Security Oversight Office told the President in a report released yesterday.
Modernizing the system is a “government-wide imperative,” the new ISOO annual report said.
But that is a familiar refrain by now. It is much the same message that was delivered with notable urgency by ISOO in last year’s annual report which found that the secrecy system is “hamstrung by old practices and outdated technology.”
The precise nature of the modernization that is needed is a subject of some disagreement. Is it a matter of improving efficiency in order to cope with expanding digital information flows? Or have the role of secrecy and the proper scope of classification changed in a fundamental way?
Whatever the goal, no identifiable progress has been made over the past year in overcoming those obsolete practices, and no new investment has been made in a technology strategy to help modernize national security information policy. In fact, ISOO’s own budget for secrecy oversight has been reduced.
Even agencies that are making use of advanced technologies such as artificial intelligence, machine learning, and predictive analytics in other areas have not considered their application to classification or declassification, ISOO said. “These technologies remain untapped in this area.”
At some point, the failure to update secrecy policy becomes a choice to let the secrecy system fail.
“We’re ringing the alarm bells as loud as we can,” said ISOO director Mark A. Bradley.