Overcoming a Culture of Secrecy: Can the United States Kick its National Security Over-Classification Habit?

The United States faces evolving threats from China, Russia, Iran, and North Korea, but a culture of overclassification of intelligence results in the routine failure to share vital information at speed and scale. In an example of bipartisanship, panelists discuss how the United States can reform its national security information policies, regulations, and laws to ensure crucial insights are shared quickly and effectively across government.

ANGELSON: Welcome, all. I’m Mark Angelson. I’m your presider for today’s on-the-record meeting. I’m a former member of the President’s Intelligence Advisory Board and current chair of the statutorily bipartisan Public Interest Declassification Board, which you’ll hear referred to today as the PIDB. And I’m a longtime member of the Council and its Membership Committee.

As you’ll hear from our panel, it’s no secret in our nation’s capital or elsewhere that the declassification system is in urgent need of attention. In short, the intelligence community—which you’ll hear referred to here today as the IC—has been overclassifying since the 1940s. The rule has been when in doubt, classify. Fast-forward; we now have eighteen elements in the intelligence community. Notwithstanding the second President Bush’s efforts to kind of get some organization around that, it hasn’t happened. And so there are now far too many standards for classification and declassification. Add to that the advent of the internet and the substitution of—substitution for ink on paper of digital communication, and lately to add insult to injury artificial intelligence, and we have before us a big problem but maybe, on account of the latter two elements, an opportunity to solve all this in the near term.

We have just the right panelists for today’s topic, all former or present officials from the three branches of the federal government steeped in intelligence policy and experience.

Carter Burwell is a partner in the eminent law firm Debevoise and Plimpton, having served in all three branches of government himself. He’s our resident expert on what the Congress might do to get this fixed, and how.

We have Congresswoman Jane Harman with us today. Welcome, Jane. It’s always a pleasure to see you.

Ezra Cohen reports directly to the CEO of Oracle. Former chair of the PIDB. Previously served in senior roles both in the White House and at DOD, all related to intelligence.

Jennifer Sims served in intelligence roles at State and on the Senate staff. She’s a widely published author with extensive experience in the academy at Georgetown and Hopkins and elsewhere.

Last but not least, this is Michael Thomas’ day job. He’s the director of ISOO, the Information Security Oversight Office, which is responsible to the president for policy implementation and oversight of the governmentwide classification system. So we’re coming for him today.

SIMS: (Laughs.)

ANGELSON: He previously served at the—at the NSC and in the ODNI—that’s National Security Council and the Office of Director of National Intelligence. I’ve been cautioned about acronyms.

We’re going to engage our panelists in discussion for half-an-hour or so and then we’re going to happily take questions from the audience.

My first question is for Michael, with others encouraged to jump in when Michael’s finished. Michael, everyone agrees that the classification system is in desperate need of reform. Tell us what this really means. And describe, please, if you would, what the system looks like today.

THOMAS: Sure. Thank you, Mark.

I think, rightly so, much of the public conversation around reforming this area has to do with the declassification portion, sort of that last mile of how do we get more information released. But I think, you know, in this group we should be prepared to talk about all the layers of the system, from how the initial guidance is created that derivative decisions about what to classify come from; how those things are marked, and how people are supported to do that accurately and precisely; how information is disseminated, and how classification affects whether information gets where it needs to go; and then, ultimately, when do we review it to assess whether it can be decontrolled, and then how do we release it to the public effectively. I think when we talk about something like overclassification, you see aspects of it at each layer in that system. So to sort of give declassification our sort of sole focus is kind of like a farmer only ever focusing on the harvest and never, like, fertilizing, and seeding, and irrigating—all the other things that lead you to have sort of a successful outcome.

ANGELSON: Others?

Ezra, please tell us which parts of the classification system have proven most difficult to reform until now and, if you would, give us your insights on why past efforts have failed to break those bottlenecks.

COHEN: Yeah. Well, first of all, thanks, Mark, and thanks to CFR for hosting us. And, obviously, we’ve got three members of the PIDB—actually, four members of the—of the PIDB here, three current members.

Look, I think the biggest thing to reform is really the culture. And for very good reasons there’s a culture of secrecy in many of these agencies, and you know, that is there for, again, very good reasons. However, we’ve gotten to a point where that culture has actually become quite self-destructive, and there is a—we have reached rock bottom in terms of the public trust in many of these institutions. A lot of that is because of the culture of secrecy. How we change that culture, right—you know, this trickles down to how the person—just, you know, the most junior person, right—when they go in to have a new job at the—at the—within the intelligence community, they are indoctrinated informally that everything should be classified. You know, you’re under time pressure; just mark it Secret/NOFORN or whatever, get it out. You know, there are a lot of people—I’m looking around the audience here—who have been in government, and I know that they’ve all—and some of the in the IC, and some of them have probably done the same thing. I certainly have.

So how do you change that, like, day-one culture that’s put into the mind of every new IC employee that’s going to stay there for thirty years? That’s really the hardest thing to change. All the other stuff, it’s essentially mechanical, right? Like, you know, we’re going to talk about today I’m sure what algorithms can do, both on the frontend and the backend, and automated classification and declassification. You know, we’re going to go through all of that. But none of that really gets at what this core issue is, which is just this culture of secrecy.

BURWELL: And to pick up on that, how the culture of secrecy is actually harming national security; that we were raised, many of us in the intelligence community or national security, understood that secrecy is critically important in protecting national security. So the challenge now is because of overclassification and a failure to declassify information. And because of this culture of secrecy, the public is now losing faith in our intelligence community. And that, ultimately, is the foundation of our national security, is the public and our government’s ability to drive common will in defense of this republic. And so the ultimate harm of secrecy is kind of destroying that, and we need to fix that.

SIMS: And underpinning that is a security paradigm where we tend to think about security first rather than competitive advantage first. And when you’ve got a security mindset, everything looks like a threat when you think—when you’re talking about declassification: What’s the damage that’s going to happen? And you know, for those of us steeped in counterintelligence and offensive counterintelligence and influence operations, you know that the release of information is as important to shaping the competitive environment as keeping secrets. In fact, keeping secrets can get you into a lot of trouble because—you know, just look at 9/11. What did we learn from 9/11? Those guys did not need to steal our secrets to do us great damage. Competitors work around your secrecy apparatus to hit you the way they can without having to steal what’s in your vaults. So if you are just thinking like a vault keeper, you’re not being competitive in the international environment.

THOMAS: Yeah. Overclassification is certainly something that doesn’t just involve information that can go out the door or not; it’s on what system it rides, where can it get to, and can it get to the place where it can do the most good when it needs to. I think that’s an underserved aspect of the conversation.

ANGELSON: Amen.

Jennifer, what specific risks do policymakers or agencies/elements of the IC cite when they defend the current level of secrecy? And is this just cultural, or is it evidence-based as well?

SIMS: OK. Some people are going to get riled by what I say—(laughs)—in answer to that in this audience perhaps and maybe on the panel. But there are very good reasons for secrecy, let’s be clear about that. Sources and methods are important to protect—lives are at stake, military strategy, military operations, the vulnerability of our troops, the supply chains for our armaments. Defense contractors working with us, their proprietary information needs to be protected. There are a lot of reasons for having a good, sound secrecy policy. Diplomats want secrecy for when they go into negotiations. They don’t want the other side to know what their bottom line is in a negotiation, and that has to be protected. That’s not sources and methods, but it’s something that provides decision advantage. And what intelligence has been about for a number of years now has been not the protection of secrets per se, but gaining decision advantage.

And the complement to that is, in a democracy, the need for transparency. And that’s the tough thing, because you need an informed public to be able to support a complicated foreign policy in a time that is not anything like the Cold War, where we all had an enemy, we knew what we’re about in the international system. Well, things are a lot more complicated now, and the public needs to understand, and is more interested in what we’re doing and why. So we have to find a way to marry up better transparency with the protection of secrets and at the same time pursue decision advantage, which is something that the State Department has done for years.

And we’ve got a semantic problem here because the State Department isn’t an intelligence agency according to our intelligence community definition. But in State Department for decades, and longer than decades—(laughs)—many, many, many decades—has gathered open-source information gray-area information through its Foreign Service reporting, and use that information in sensitive negotiations. And what we’ve done is undercut that agency over the years by reducing Foreign Service reporting staff and the reporting function and all the rest of it. Those were the people who understood how to turn protected intelligence into decision advantage overseas, and do influence operations with other governments.

So we need three things. We need the protection of sources and methods and deep secrets. We need transparency with the American public. And we need to pursue the culture of a decision advantage, which is the gathering of sensitive information through diplomatic channels and using it through diplomatic channels to do global influence.

COHEN: Can I—can I?

ANGELSON: You may.

COHEN: Yeah. So, I mean, I think, you know, the virtues that Jennifer speaks to are very important. I think the general problem is that the American people, you know, speaking from the public interest, feel that this system has been abused and they don’t trust this, right? All of those, you know, things that Jennifer spoke about, those are basically—those are lines in what we call security classification guides. Those are the justifications to make something classified. And we know—I know, the board knows, it’s publicly known—that those justifications are abused and they’re applied to information that they don’t necessarily apply to. There’s no way to check it. There’s no way to monitor that.

And you know, speaking of diplomatic cables, look, the cables that were leaked by—you know, through Wikileaks, you know, obviously, that was a—you know, there was damage that was done by that, but I think a lot of the public looks at that and says: Wow, there is—there is actually information in here that the public should know about. There are things in here that, you know, probably aren’t secret. They’re embarrassing, but they’re not—they shouldn’t be classified. We should—we should see that. So, you know, there’s this balance.

I think we really—part of the thing is, is that if we want to actually protect the secrets that need to be protected, we have to reestablish trust with the American people that the system is not overclassifying. And there’s a lot of work to do to get there.

THOMAS: Yeah. I think the—speaking to your cultural issues that you raised, Ezra, you know, in the intelligence community or in the DOD, classified holders, you know, we talk about transparency and people—

Q: Can’t hear you.

THOMAS: Can’t hear me?

COHEN: I think the mic’s not on.

THOMAS: Is the mic off? It’s on?

All to say that people hear the word “transparency” and they think, you’ve giving away my secrets. But if you—if you look up the word, it actually means to make something visible without distortion. And I think that’s part of this cultural aspect, is that the people who use this information day in, day out and live in that world don’t realize that the way they are handling it often creates a distortion in the way the public understands what we do. And in times when it is crucial for the public to trust a conclusion that comes from the intelligence community that is in the voice of the president or in the voice of our diplomats, we’ve eroded that. And so to give greater visibility into how decisions get made and the expertise that drives them is crucially important. But it doesn’t mean giving away the farm; it just means making sure that you’re taking care to present the information without distortion when you provide it to the public.

COHEN: And you know, when it takes forty or fifty years to release information about a historical event, you know, nobody has absolutely any faith, including people that are in the government, in the fact that the information is not going to be distorted when it’s finally released.

ANGELSON: Ezra, tell us what’s at stake here, please. What are the costs of overclassification?

COHEN: Well, look, I mean, there’s a lot of costs. There’s, obviously, the cost to—you almost knocked your water. (Laughter.) There’s, obviously—there’s, obviously, a cost to the, like, health of society, right? Like, this is what we’re talking about. We live in a democratic society. We don’t live in a security state. We don’t want to be in a security state. So there’s a—there’s a serious cost there when you have over-secrecy. I think that’s the highest cost.

Then there’s the dollar value, right? I mean, the reality is classified information is way more expensive to store per page, and you know, there’s a cost to that.

There’s a cost to our cooperation with our closest allies, right? I mean, if we’re putting a NOFORN stamp on everything, it makes it very hard absent, you know, extraordinary intervention—which we saw the last administration do with Ukraine—absent extraordinary intervention from, really, the top, right—top-down intervention—it slows down sharing and partnering with our key allies.

And so I’d say that those—I would list those as the three major costs.

THOMAS: Thank you. Thank you. Thank you.

ANGELSON: Carter, legislation and beyond legislation, what oversight or budgetary tools could Congress use to facilitate or, in fact, to compel reform in the classification practice?

BURWELL: You know, we did have—thanks to Senator Cornyn and Senator Wyden, a few years ago we did have some legislative activity, the Senate—Senator Warner—to try and encourage some reforms. Sadly, some of that—some of that effort still hasn’t been implemented, some of the reports that were required by that legislation. So there are mechanisms to do that. In fact, the board—our PIDB is supposed to do that, and we’ve been operating, you know, skeletally for years and years now without a budget. So there are methods to raise awareness. And again, back to what Ezra said, you know, grateful for you guys in putting on this event and being here to become aware that there is significant public interest in declassification.

But Congress can pass legislation. It’s pretty hard to do so, especially over the last six weeks or so when they were shut down and not present. You know, the Senate Intelligence Committee and HPSCI could have hearings on the subject. I think HPSCI had a hearing a few years ago. It’s not necessarily the most salacious or most interesting topic. These are very weedy things about how to improve and limit classification decisions on the front side, and the detailed process that Michael and ISOO and the Archives do in trying to encourage and promote declassification. These processes are very challenging, and opaque, and obtuse, but they’re important. And I think the government—sorry, the people are now demanding it because, as Ezra said, the people are losing faith in their government.

SIMS: I think this is a really important point. But another important point, if we’re going to get practical, is that to overhaul our classification system you have to get the intelligence and counterintelligence community onboard. And you’re not going to get their attention with transparency arguments because that runs counter to everything they’ve been trained for. It runs counter to what they’ve achieved in the field through sensitive operations. It feels excessively risky to them, and as not part of their mission statement.

Actually to make progress on transparency you have to bind the community together: intelligence, counterintelligence, the FBI, and the public—the people who advocate for transparency. It has to be a whole-of-government approach. And that you only get by being able to hold in your head at the same time that the risks and dangers associated with losing secrets, which we are constantly generating still, hold that in your head at the same time as you hold in your head the notion that the whole point of having secrets is to use them and to gain certain advantages from using them. And it requires discernment to balance the two—when do you keep the secret, when do you divulge it to gain some kind of international advantage, and when do you release it to the public for purposes of transparency.

That is a complicated three-way balancing act that requires everybody involved in the process to think in transcendent terms about it. And if we bifurcate it into transparency over here to get the public on our side; or secrets over here—don’t mess with our classification system, because you mess with that and we could lose control of information; and the counterintelligence community, which is all about, you know, they’re trying to get information on the Brooklyn Bridge, let’s classify that, you know, because they want to blow it up; there is a dynamic going on day in and day out with people trying to do their best to do good work in these institutions according to the mission statement they’ve got and the training they’ve received. And somehow we have to weave all of those together to make progress. Otherwise you’re going to have roadblocks thrown up all over the place.

COHEN: Can I—can I—

SIMS: Yeah.

COHEN: Can I just—(inaudible). Yeah. I mean, I think—again, I think—I think Jennifer’s exactly right. But I think we are sort of past that point in the sense that the—there are very disruptive—you know, bureaucratically disruptive—I think from the perspective of a bureaucrat there are things that are disruptive that are happening now because they went too far down the road. The pendulum swung way too far, right? The whole idea of public transparency, it’s supposed to be a balancing act between the public interest and the need to keep the information classified. And it swung way too far in the direction away from the public interest, and we are now seeing a reaction to that.

Look, what the board is—just to pick up on what Carter said, the board is not advocating generally for the, you know, extraordinary mass declassification of documents related to historical events. Yes, that is something that is very important and we totally support that. You know, we’re advocating right now for the 9/11 Commission records to be made public, extremely important ahead of the twenty-fifth anniversary when these are going to meet their mandatory declassification review date. We were very early on the JFK stuff, you know, thanks to John Powers, and you know, we are advocating for that. But what we’re really trying to get at here—and this is—this is the key thing—the hard part is that structural, core systematic change. That’s really what we’re trying to get at as a board.

SIMS: I—

COHEN: And that requires—sorry, one last thing.

SIMS: Sorry.

COHEN: That requires congressional action and funding. Sorry.

ANGELSON: No, no. No apology’s in order.

AI, Michael. Talk to us about AI—the benefits, the risks, the opportunities.

THOMAS: Yeah. I think, you know, for years we’ve had a problem with this tsunami of digital records, right, across all government records, not just classified records. But we generate so much more in the digital world, and there are not enough humans that will ever be assigned to these tasks to make our way through it. So one of the challenges we have getting information out is just having not enough people to even look at it, right?

So we are living in this world of AI automation now, and we can see clearly how those tools should be able to assist in that process. And they are. There are really bright spots happening all across government. You could point to several. You can read about several on the PIDB blog. They recently did some visits out to check these projects out.

So we’re in a world where these things are possible, and problems that used to be enormous can now be solved. And I think the potential there is going to fulfill some long-term needs that have been asked for by people like you. People want to know how many—how many documents they’re creating, how many derivative decisions are you making, what are the different metrics that matter in classified information. And the answer is right now there’s no way to collect them. It’s my job; I can tell you quite clearly there’s no way to collect them. But with these tools you have an opportunity to build from the start the type of analytic insights that you would want to derive about the system to really describe its health to you to allow policymakers the information they need to have, heads of agencies the information they need to have to make good and better decisions for real risk analysis about the system itself.

Secondarily, you have an opportunity for a drive towards centralization and insight into how the system is operating so that you really can have meaningful oversight. That’s one of the things that keeps me up at night, is that how do I oversee an algorithm? Do I have the workforce that enables us to do that? You know, we’re going to reach a place very soon where we’re going to have to move beyond sort of the—the sort of artisanal barista handcrafted model of classification, have to trust some layer of machine assistance to do this, not just on the declassification end but from the frontend onward. When you log in to craft a Word document, an email, you could probably expect a little bit of assistance at some point in, like, a lot of people classify it this way; you’ve classified it that way. So we’re going to have to get used to thinking about these tools in every aspect of the system, not just where we’re seeing most of the action right now, which is at the tail end.

SIMS: Could I just add on?

ANGELSON: You may. Yes, please.

SIMS: An important point on this is that an argument has been made, at least in academic circles, that AI in the hands of adversaries will crunch what we release through declassification measures and reverse engineer in such a way that we can—they can figure out what was deleted. I think that’s an erroneous argument to make. It should not slow down the application of AI.

There are a lot of ways around that. You can use bracketed deletions, other things. But it is—you know, other governments are going to be doing exactly what we’re doing, and that will fuel their capabilities including causing counterintelligence problems for us. But that should not slow us down. We can talk more about that later, but—

ANGELSON: Carter, there has been some progress, maybe. Tell us—give us some examples, maybe, or say whatever you want about what good looks like in today’s system.

BURWELL: Well, I think they are—the intelligence community is making some progress using AI. That is a good—that is a good-news story, and you can read more about that effort to the extent that we can talk about it publicly on our blog. We’ve done some work in that space. And that is good news because—(laughs)—like Michael’s saying, there’s not enough money in the world to deal with this problem if we don’t use LLMs to address it. So that’s good news.

I think the fact that Congress got together and tried to focus some attention on this. You know, Avril Haines made a great speech—we had an event down in Austin a couple years ago and Avril Haines made a great speech on her intended effort to address these issues publicly and within the intelligence community. Sadly, I think the politics of classified documents kind of got in the way on that issue related to, I guess, current and former presidents. Kind of, you know, caused all of that effort to stop as the parties went back and tried to operationalize classified information in support of their own agendas, which was really disappointing to see because I think we had some real positive momentum there—bipartisan, apolitical momentum.

So there are, you know, some bright spots. I think the current DNI is active in the space of declassification. And you know, to see senior officials want to do that—obviously, this Epstein bill that just came out, I was actually a little disappointed. I actually read it, and it says declassify—or, you know, make public everything except all the classified information—

SIMS: (Laughs.)

BURWELL: —which God only knows what that might be. But you know, we could take care of that, certainly to the extent that there’s public interest in the subject. So there’s—there are paths forward if we just focus on this.

COHEN: Yeah. I mean, you know, I’ll just say that I do think that DNI Gabbard has been really good at representing the public interest, and that’s important. That’s something that I do feel has been missing.

You know, regarding the structural changes and really LLMs, to come back to the initial question, what’s happening now is you have a dozen or so independent separate efforts—maybe even more now—to develop some sort of automated classification-declassification system within the U.S. government. There’s a lot of overlap. It’s a little bit of a Venn—very complex Venn diagram. You know, some are doing these features, others doing those features. But I can tell you this, it is extremely wasteful because there is a huge amount of duplication of effort going on.

Congress tried to solve this by putting in the law that the federal CIO essentially would be responsible for the one central program to develop an automated system to deal with this problem across the U.S. government. It was in the law, President Biden signed it into effect, and then I think they put it in a desk drawer and didn’t do a thing with it.

SIMS: (Laughs.)

COHEN: So, you know, we really desperately need, even to this day, a focused, centralized effort to address this problem across the whole government because, you know, as we know—and the public has become very aware of this now—but there are very few number of classified documents that only have one originating agency. There’s so many equities in each of these documents that, you know, you’ve got to have one central agreed-upon way automated to kind of adjudicate whether or not that document can be released to the public.

ANGELSON: Thank you.

So I’m going to ask one last question and I would be grateful if each of you would respond, and then we will turn the questioning over to the members of the audience. Starting with Jennifer, if you could ensure the adoption of one concrete achievable reform for current—for the current administration to adopt, what would it be? And what would success look like, say, five years from now?

SIMS: OK. I’m going to have one reform with two parts. (Laughs.)

The reform, first of all, I think that the Public Interest Declassification Board—and this is not self-serving because I’m not serving on it now—it needs to be better funded and it needs to have a staff. And it needs significant resources put against the transparency issue. We should not ask the Public Interest Declassification Board to adjudicate what I’ve talked about as and others have talked about as intelligence diplomacy, if you will—that is, the real-time ramped-up process of demarching, you know, of getting documents released as we did before the war in Ukraine out into the public space so people can understand what’s really happening.

One of the downsides, of course, is the lack of trust in the—from the public side in government institutions makes this hard to do. But I think that—and we have an assistant—former assistant secretary INR here in the audience, Toby Gati—I think we need perhaps the assistant secretary INR to have—to chair a board with as co-chair somebody from DOD on the policy side to work with the intelligence community through the ODNI on intelligence diplomacy.

I am not in favor of a freestanding national open-source intelligence agency, creating some kind of new agency. I think my comments before about the State Department mission has been in the past and is underfunded for now, it needs to be ramped up again. Foreign Service reporting needs to be out there. They’ve been doing intelligence diplomacy for eons. We just need to modernize it and make it more ramped up, high powered, with more Foreign Service officers working on it, and probably chaired by INR.

ANGELSON: Thank you.

Mr. Cohen, sir, same question. One thing, what would you do?

SIMS: (Laughs.)

COHEN: OK. Yeah, I think that the—that the most achievable thing—because that’s very important, is to recommend the most achievable thing—is to create this centralized effort for a(n) AI algorithm. And also very important in that is that there’s more classified information that exists outside of the intelligence community than exists within the intelligence community. So when you start looking at Department of Energy, the State Department, DOD, the volume of classified information is extreme, and that’s not part of the IC. So you’ve got to have this sort of neutral entity that exists outside of the IC but also the other agencies that creates this centralized, whole-of-government effort to address the AI algorithm problem.

ANGELSON: Mr. Burwell of Virginia.

BURWELL: (Laughs.) Put declassification reform into the White House.

ANGELSON: Fair enough.

Last word on this topic is yours, Michael.

THOMAS: Well, I’m of a like mind with Ezra, but I’ll be even more tactical. As we approach declassification in real time, if anyone has dealt with this, there is the issue of agencies that have—documents representing multiple equities. Right now this is a pass the folder to the next office, one at a time. Have you counted the number of intelligence elements we have? We have a lot of them. And so this is a very slow process.

Something as simple as a cloud instance where agencies could collaborate together cross-platform to review documents concurrently and collaboratively would speed the process. I think that also creates a beachhead for testing out the algorithms, a centralized way of looking at how to do this work, as well as a place to share guidance, as well as a vantage point from which oversight could look in and analyze the system. So I would advocate for a very achievable cloud instance for classified review.

COHEN: OK. And I want—you know, I heard some gasps when Carter made his comments. (Laughter.) So, first of all, people should understand that historically classification management reform has been at the White House. This is a thing that really is this—really, the sole domain of the president of the United States. It’s an executive power. The Supreme Court has reaffirmed that repeatedly. And the reality is that’s where the money—the money—the dollar decisions are made.

And this is a problem that needs a focused, funded effort that isn’t going to run into—you know, and, look, I was at that kind of, like, you know, fence between DOD and IC, and I can tell you the two don’t get along when it comes to this issue. So we really do need some new—we need a neutral ground here.

BURWELL: A decider.

COHEN: Yeah. A decider, exactly.

SIMS: But getting a neutral ground—just one second—getting a neutral ground doesn’t mean that you’re going to be able to operationalize it. The White House doesn’t have operational authority. It is in the agencies, and they can throw up all kinds of roadblocks. You can get neutered in the White House. If you’ve got the—if you’ve got an agency head going in to see the president in the Oval Office, somebody on the staff of the White House trying to argue declassification issues is going to get run over like—

COHEN: It’s all about the money.

SIMS: Yeah.

ANGELSON: It’s not the money.

COHEN: The funding.

ANGELSON: It’s the money. (Laughter.)

OK. Thank you all. I’m going to turn now to the audience for questions. Please raise your hands. I see five hands, but not—OK, I’m going to get to all of you, I promise. Wait to be recognized. I think a microphone is going to be brought to you, yes? And please start by stating your name and affiliation.

The first question will be from Henry Sepulski (ph), who is the person in the private sector who has done—please come to the front with the microphone, right here—who is the person in the private sector who, with his organization, has done the most thus far to get us to such progress as we’ve had, and particularly in respect of appropriations from the Congress, et cetera. So Mr. Sepulski (ph), sir, the first question is yours.

Q: I’m afraid your introduction tells us more about what trouble we’re in than what I’ve done. (Laughter.) I think we have done squat.

The board has been making the arguments about algorithms now for, I don’t know, six years? OK. You’re not—

ANGELSON: We need a—we need a question.

Q: You’re going to get a question. (Laughter.) You’re going to get a question you won’t want to hear. (Laughter.)

You guys talk about intelligence community. Give me a break. The problem is in acquisition for sure, and you haven’t said a word about that. There is something that I asked the board in June of last year called controlled unclassified information. Not a whisper from any of you. Cost 4 billion (dollars) a year to implement—4 billion (dollars)—and it doesn’t add any value.

BURWELL: What is it?

Q: Well, that you don’t know is why you knew to call me.

BURWELL: No, no, I do, but tell everybody what it is.

Q: Controlled unclassified information is a management tool to control information that is unclassified such that you will not be able to get access to it.

ANGELSON: OK. Henry (sp), we need a question, please.

Q: The question is, when I approached the board in June of last year, we don’t want to touch that with a barge pole. Now, in light of the desire to have Homeland Security narratives that are compelling, and to push the Golden Dome—which is under gag rule, by the way—has the board changed its position at all?

ANGELSON: OK. So the board has changed—please. Thank you very much. The board has changed its composition in the first instance, OK.

Q: Oh, I understand that. That was the reason I’m asking.

ANGELSON: OK. So, Ezra, would you—would you—

COHEN: Yeah. Yeah, OK.

So, first of all, Henry (sp), before I respond to your question, and I’m going to respond, I really do want to thank you for everything you’ve done. And for everybody here, if you haven’t read Henry’s (sp)—the NPEC’s report on CUI, and if you care about this topic I highly recommend reading it. It is very detailed and makes a lot of good points.

OK. So first of all, you know, this is a very, very different board. We’re very forward leaning now. We went through a period where that was not the case. There was clearly direction to make sure that the board did nothing. That has changed.

BURWELL: Thanks to our new chair.

COHEN: Thanks to our new chair and so—

BURWELL: And vice chairs.

COHEN: Yeah.

ANGELSON: And vice chairs, please.

COHEN: So that problem has been resolved. You know, we obviously have a lot of—there’s a lot of demands on the board. I think it’s something that, you know, we need to—we should take up. I really think we should look into it.

Now, so, you know, I can’t, like, vote for everybody on the board. I’m telling you publicly now my position is I think it’s something we should do, OK?

Now, I just want to say one other thing on CUI. I actually believe that the reason for CUI comes from a good place. Look, we’ve got—we know we’ve got the CCP and the Russians and, you know, some other countries that are actively trying to steal and take advantage of the fact that we have an open society, that we have—we rely on a commercial capitalist base, defense industrial base.

There’s a lot of sensitive things being done there. Those industries all have a lot of suppliers from, you know, small machine shops that make screws to, you know, really big shops like Lockheed Martin, right? And it’s a whole range of things.

And so I think that the intent of that—especially when you have interagency coordination on these programs across multiple agencies, the intent was let’s come up with one way to protect this sort of sensitive information that’s not classified in one uniform way across the government.

Now, it’s been fifteen years since the EO was signed to do this and it’s still not implemented. So, obviously, there’s a serious problem, OK? And every agency has implemented it differently and a lot of agencies are still using FOUO and SB—you know, sensitive but unclassified and law enforcement sensitive and all these other markings.

So, obviously, the implementation has not gone well so there needs to be a real look at, you know, what went wrong while also accepting the fact that we do have adversaries that are very actively targeting these non-intel, highly sophisticated classified parts of the U.S. government and defense industrial base and we got to figure that out.

So I don’t think it should be nothing. I don’t think the answer is to, like, go to nothing. But, you know, there’s some middle ground there, and also it’s got to be implemented quicker than fifteen years.

SIMS: Can I just ask a question of Ezra?

ANGELSON: Yes. Yes, you can.

SIMS: Very quick. Very quick.

ANGELSON: But not just—but not just yet.

SIMS: OK. OK. (Laughs.)

ANGELSON: The gentleman seated next to Meaghan Fulco. Meaghan Fulco, managing director of the Council, is the person at the Council who put today’s event together. Thank you, Meaghan.

Please identify yourself.

Q: It’s a great event. My name’s Jonathan Paris. I was a Middle East fellow at the Council on Foreign Relations in the ’90s. I moved to London in 2001 and became a consultant to the Office of Net Assessment and then a NIC associate for about fifteen years.

My question is, since I never really had security clearance I was able to do maybe three dozen reports and studies based on open source and I wanted to know, since I’m writing mostly about futures of countries outside or phenomena outside the United States, what exactly is the advantage—what exactly is my disadvantage of not having access to classified information? Thank you.

ANGELSON: Michael, you want to do that?

THOMAS: Well, first of all, it’s the truth about aliens. No, I’m kidding. (Laughter.)

No, I think what is interesting about your question—and at the DNI I work with the NIC a lot and work with NIC associates a fair amount on Global Trends project and things like that—I think one of the—and I saw this most clearly in the MH-117 shoot down and the public discussion about that—that there was a lot in the public sphere, and as the intelligence community briefed the media and interested parties outside of government there was a question of, like, why won’t you say what we think we see on Twitter at the time.

And the answer was in part because the way that we know it and can validate what you think you see is what makes it special, and speaking to decision advantage, part of maintaining decision advantage is maintaining the capabilities that let you keep knowing those things.

And so I think maybe more now than ever more is accessible in the public sphere that is similar to what we know in intelligence circles. But in intelligence circles you have this extra layer of validity and sourcing that gives you, you know, context and assuredness where, you know, 80 percent might not be enough to do certain things that you need to do in the context of a conflict.

So that’s the first thing that comes to my mind.

ANGELSON: Anybody else?

Congresswoman Jane Harman in the front row here, please—the lady in orange.

Q: So this is very interesting and how little has changed in the years since I fled Congress. (Laughter.) I should also add that I was fired from PIAB recently, so I guess I have a very good record here.

But there’s a dimension that you didn’t talk about that I think is really relevant. After I served on HPSCI I chaired the Intel Subcommittee of Homeland, and we focused on vertical intelligence sharing from this marvelous collection of agencies to the cop on the beat, et cetera. And that’s not just relevant for Homeland security purposes, it’s also relevant for national security purposes, and what was pretty terrifying was things would start out in D.C. being over classified and they would somehow get to somewhere where there was no ability to see them.

And so I remember once there was a ricin scare, which could come up anywhere, in Las Vegas and somehow it was classified what ricin was. Don’t ask how lame-o that was. But the point is how do you know what to look for if you have no idea what it is? And this kept going on and we kept making various changes and nothing worked. So my question is has anything changed in fifteen years.

ANGELSON: Volunteer?

BURWELL: I don’t think so. (Laughter.)

Having worked counterterrorism cases for the Department of Justice involving ISIS-inspired teenagers or, you know, al-Qaida threats inside the U.S., no. I mean, some of that is, you know, understandable like trying—you know, thank goodness we were trying to push information down, you know, and I think the challenges of doing that.

But I think where we are now is the folks in Nebraska don’t feel like they’re being told and how can we share more information. That’s—to answer your question I think is—

ANGELSON: Jennifer, please.

SIMS: Thank you. I think this is a really important issue.

One of the hallmarks of a very successful intelligence capacity is the ability to get decisive information to the people who need it, the decision makers ,and what we’re finding more and more with the change in the international system that that’s at the state and local level. It’s your cop on the beat.

And after 9/11 we found out that there were some police departments like in Los Angeles that had done a lot of work on this and had worked with the FBI and developed kind of centers for handing off information with one another.

Well, that all sort of fell away. In fact, we’re worse off today, I think, and I don’t know. If there’s somebody in the audience who knows better, speak up.

But from my understanding, we’re not as good as we were after 9/11 when we had that critical moment where we realized that the DCI—then DCI back then—knocking on the doors of the White House wasn’t getting the information to the airlines that needed to know that there were dangerous people trying to get on their planes who had practiced flying craft but not landing them.

I mean, there is also in the public a lot of information the government needs to know. So it’s not just, you know, we’re not good at getting information out. We’re not good at the interchange necessary for a democracy to have a vibrant intelligence capability.

So I think you’ve really put—Congresswoman, you’ve really put your finger on a very important issue. The decision advantage isn’t just about federal decision making. It’s state and local, and we have to get better at that. And that’s not just a transparency issue either what is unclassified now. It’s getting tear line information out.

ANGELSON: There were more questions—

COHEN: I just want—I want to—

ANGELSON: Please, quickly.

COHEN: Can I respond to—

ANGELSON: Yes, you may.

COHEN: OK. So the—look, what has changed—what’s changed is that, you know, we now have basically full transparency on the JFK records. We had information which through presidential instruction was released, you know, before the Russian attack on Ukraine.

You know, we’re getting information on MLK. We’re, hopefully, going to get 9/11. But those are all extraordinary things. What has not changed and what desperately needs to change, given the way things are going in the Pacific, is that—you know, putting on my, you know, hat as the undersecretary of defense for intelligence, when I have information coming off a satellite—a very sensitive satellite—I don’t have time to have a foreign disclosure officer sit there and tell me if I can release that stream of information, you know, in combat to a partner in the Pacific.

I need that to be, like, real-time stream of data and it’s automatically essentially cleansed of the information that makes it of a higher classification so they can immediately get out to that partnership in the Pacific so that they can take action on that threat that the very sensitive satellite saw.

That does not exist. It’s sort of, like, that’s like the defense equivalent of, I think, the problem you’re talking about in the homeland context. But I think that that has not changed and, again, that ties directly into this idea that we need an automated system to deal with these things.

ANGELSON: The lady in the—on the aisle here in the third row, please, and then after her the gentleman in the middle two rows behind.

Q: Thank you. My name is Toby Gati. I was the assistant secretary for INR for three-and-a-half years.

And the role was crucial of INR and it wasn’t because it was secrets. It was because of judgments, and maybe judgments are secrets if you don’t want to hear the judgment but in effect I’m really sorry to see that INR is no longer around to say, guys, you have the facts but you’re putting them together in a way that the puzzle is never going to be completed.

So my question is that there are some benefits of classification. One of them is that the intelligence committees are not as politicized as the rest of Congress. Now, that may not be true anymore but it was definitely true at that time when INR was even—

ANGELSON: Still true.

Q: —part of the yearly brief to Congress because they wanted judgments and not just information.

So my question is the following. I didn’t write for declassification, I wrote for classification and I wrote for the secretary personally a lot. Twenty-five years have passed. Most of the people because of good medicine are still alive. They are not happy to read about what I wrote about them to the secretary.

So how do you deal with that? Because it’s a real issue, and it’s much easier to do classification with three letters, CYA, if you all know what that means, and that would be the best classification.

So how do you deal with self-censorship which is what’s going to happen if you declassify some of this stuff without going to the people who actually wrote it or thought about it?

And then just a contemporary question. How do you judge the chat by the secretary of defense on Signal and do you think of that as, well, you know, we’re glad we got the information out, or do you say, whoa, guys, you know, why are we doing that?

And if we had such good intel on Ukraine why did we have a policy that didn’t enable us to do more to help them since we knew exactly what the Russians were going to do and why are they going to do it, and it didn’t translate into policy.

ANGELSON: Thank you. Thank you. Thank you very much.

On your point about cooperation among members on both sides of the aisle, I’m delighted to report that the board’s experience is such that between Senator Cornyn and Senator Wyden and particularly with their staffs there is remarkable old-fashioned bipartisan cooperation and it works, and it’s not broken and it doesn’t need fixing. So there’s that.

Now, who’s going to take on—oh, Ezra?

COHEN: No. I mean, I’ll just—I’m going to answer all—I mean, just sort of answer this generally.

ANGELSON: Please.

COHEN: Look, first of all, the job of the board is to represent the public interest. I understand very well from personal experience the need to be able to write freely within the intelligence community.

I totally get it. Totally get it. And I understand how analysts feel about it. I understand how collectors feel about it. I understand how our operators feel about it. Totally get it.

I am telling you that one of the reasons INR and a lot of these organizations are going through a lot of pain today is because that—the public interest was not respected. OK. So there has to be a change in the way that we’re thinking about this, and there is a proper middle ground. Again, the issue is we went too far away from the middle ground. That’s why we’re in the situation we’re in today.

And I’ll tell you what, when it takes—I’m talking about the JFK thing now, OK? There was a lot of people in the State Department, because I dealt with it when I was at the NSC, and a lot of people in the State Department and in other places in the intelligence community that said if we release these documents there’s going to be catastrophic harm to our diplomatic relations.

All the documents are released. There’s been no catastrophic harm. Cry wolf. That’s why we’ve got a problem. Got to be very careful, OK?

So, now, on these other things, you know, I appreciate you asking a question about both administrations. Just like Mark said, this board is truly one of the few places in Washington that is in a non—really non-political.

So I’m not—we’re not going to get into these current things. This is just not the place for us. It’s not in anybody’s interest, OK?

SIMS: Can I just add on a bit to that?

ANGELSON: Sure.

SIMS: I agree with a lot of what Ezra said but I want to put in another comment here.

We are in a world of influence operations now. If you release privileged conversations among people who are doing their best to do influence not only within their bureaucracies but also with foreign leaders, you are going to have less frank conversations, less rich conversations. People will be in CYA mode all the time and you will get worse decision making.

So transparency in the public interest has to be balanced by the public interest in effective decision making within the government that taxpayers pay for.

COHEN: But, you know, Jennifer, it’s one of these things that’s, like, you abuse it you lose it and—

SIMS: Who—

COHEN: Sorry, just one second.

Now, I agree with what you’re saying. We do need secrecy. I started off by saying that. But the issue is we—and I think the JFK thing I know it sounds—might sound trivial but we cannot underestimate what the behavior around the JFK and the release of those records for—I mean, I don’t know how many, John, you wanted to say how many years it was. I mean, I don’t know how many years it was it took to get those out. Forty years? That caused a lot of damage, that one—the behavior around that one collection of records.

SIMS: Got it. That’s not what I was—

COHEN: OK.

ANGELSON: OK. The gentleman in the center, first row from the back.

Q: John Powers, acting historian at the Department of State, and because I’m a public servant I will preface this with the views I express are my own. The question I say is my own and not that of the government.

And I’d love to get your reaction to kind of my take on Mark’s question, which is using a pilot project to eliminate agency equity so there is no agency equity and protection to force the culture change starting with the 9/11 Commission records and starting with something historical, to put an asterisk on something Jennifer said, is that secrecy does wane.

Sources and methods wanes over time just as conversations wane over time. Starting with something old by ending the equity so you have to get everybody in the room, come up with a single guide of what should be classified, what shouldn’t, and build off of that.

And then you can see, hopefully, a successful project that can build its way forward. But starting with the Afghanistan commission records, starting with the 9/11 Commission records that’s historical, that also I worry about with another build the transparency back from the JFK kind of disaster.

Thanks.

COHEN: Well, I want to call out one thing that John said which is super important, and I neglected—we neglected to mention this and it’s actually something that Henry’s (sp) also spoken a lot about publicly, which is that you can’t really go full throttle with the algorithms until you’ve got one security classification guide across the whole government because right now you’ve got—every agency has its own set of guides, plural.

There can be hundreds of guides in different agencies, you know, and there’s a lot of conflicts between them, and so I think when you look across the government I wouldn’t be surprised—Michael probably knows the exact number—I wouldn’t be surprised if there aren’t about a thousand security classification guides across the whole government and there can actually be conflicts in there.

So how do you train the algorithm on what the rules are when the rules are all in conflict with each other? The National Geospatial-Intelligence Agency has done a phenomenal job of creating one guide and everybody should be following the model that NGA set of how to get down to one guide, and once every agency has one guide then we need to apply that same model to create one guide for the whole U.S. government. A very important thing, John.

ANGELSON: If you remember nothing else about today’s meeting please remember what—John’s question to what Ezra just said. One guide.

Next question, please. Gentleman here in the third row and followed by the gentleman on the aisle on the far left.

Sir?

Q: Jon Rosenwasser, Duke University. It’s great to see you all up here today.

COHEN: The volume’s not on.

Q: Jon Rosenwasser, Duke University.

I want to first—I thank the Council for their leadership on this topic. This is not a topic which I expected the Council to take up and it’s great to have it on primetime agenda and I want to—I thank all of you for your leadership on this topic.

I agree with Ezra absolutely that it’s a cultural issue and it’s a structural issue so I have two questions about that I wanted to raise. One is whether and how ISOO should be at the archives or should be separated out. I’m not going to ask ISOO to answer that question but I would like to welcome the others to their commentary on that.

I know Carter had some observations on that because my first question of whether you can have a government wide policy be implemented out of the archives is potentially, like, problematic.

My second is congressional oversight. Right now the ones who have oversight of this primary jurisdiction it’s not HPSCI and SSCI, it’s Government Affairs, and it’s hard to get anything out of the Congress. Do we need to have a select committee on classification in order to look at this topic? Thank you.

ANGELSON: Thank you very much.

Carter, you want to start?

BURWELL: Well, it is a discussion that we’ve had amongst the board for some time is whether ISOO needs to be moved up to the White House—maybe to OMB itself, maybe its own separate office there.

But the challenge of doing this from archives is significant and, you know, I think maybe you’re seeing the results. You know, if you were going to design a system to not have results this is one you may—you might come up with this in a way. You know, not suggesting that would be a grand conspiracy.

So, but look, like, change would be helpful. Energy would be helpful, back to the point that PIDB’s been operating without a budget, you know, staffed dutifully by folks at ISOO and the archives through very challenging periods of time so, you know, and as a credit to the leadership there and the effort there.

But I think something should change and, right, you know, doing the same thing over and over again with no result is the definition of crazy or something.

ANGELSON: Ezra?

COHEN: Oh, I wasn’t going to say anything. Do you want me to say something? (Laughter.) For once. (Laughter.) Yeah, look, I—you know, honestly, Carter did a perfect job answering that question, so.

ANGELSON: Michael, you may not want to say anything, but you have an opportunity to speak if you’d like.

THOMAS: No, I just think one of the unique things about having an entity like ISOO which oversees the classification system as well as the controlled unclassified situation—I’m very faithful, Henry (sp), because they made me head of the church—is that it is a neutral player relative to the various interests in the IC and the DOD.

And so I agree with what my colleagues have said here that it has not always seemed like it was a priority of the organization. I agree with them that that has changed. I feel very supported in my role there, and it takes time to reconstitute after a long period of drought. But there are some benefits in that.

COHEN: Yeah, and I want to say this is—I mean, just this is a very objective observation. The focus on this issue at the National Archives is getting more attention now than it has in the past, you know, ten years, at least that I’m aware of. So there’s a lot of focus on this issue right now.

ANGELSON: And then kudos to Michael Thomas, who is a breath of fresh air, and we have high hopes for you and we hope that you don’t move to the White House anytime soon. But you may.

Gentleman on the aisle over here.

Q: Hello. Brian Babcock-Lumish, Institute for the Study of War.

I’m curious about whether we ever have a—excuse me, ever have a chance of kicking this over classification habit when those marking intelligence for classification are, in a sense, responding to the perverse incentive that policymakers and consumers of that intelligence have the implicit bias that more classified equals better and more reliable and these sorts of things.

THOMAS: Yeah. I think, you know, Ezra, you mentioned the CoNGA guide, the NGA’s consolidated classification guide system. There are two really important facets of that that I don’t think get enough attention.

One is the inclusion of additional data about each classified line item—we call them, like, enhancement statements—that tells you exactly what about that information is classified so you can make a more precise description, and also you can talk about it in an unclassified way more effectively.

You can convey the insight without having to expose the classified aspect, which means you can share the information better. I think people at NGA are much more enabled to do that based on the reconstitution of this guide.

The second, and this maybe gets to sort of a procedural element of what John talked about, is that we know if people use classification guides that they’re supposed to be updated on a regular basis, but this might happen once a decade realistically. It’s an enormous undertaking and very small staff.

At CoNGA they’ve instituted a process where there’s a feedback mechanism where guides get updated in almost real time. Questions come in every month. They’re adjudicated very rapidly by a small panel that is empowered to make those sorts of decisions, and the guide is updated because it’s available digitally.

I think these are dynamics that we can see that will enable people at the front end making those initial derivative decisions to make better, more precise decisions and it’ll have a strong impact in cutting back on over classification at the point you’re talking about.

ANGELSON: Amen.

So Council meetings end promptly on time. We have two minutes left.

Dr. Allan Goodman in the front row and the gentleman right there, if you would, please. So two quick questions. I’m going to ask you to just ask them serially and then we’ll give you an answer.

Q: Thank you very much, Mr. Chairman.

Is what you’re discussing a unique American problem or would our allies be having the same panel?

ANGELSON: And then—so hold that question.

Q: I’m glad he hit allies, because now I will hit the industry part, and that is—

ANGELSON: Your name and affiliation?

Q: Oh, yeah. Sorry. John Galer from the Aerospace Corporation.

But as we see a push, particularly within the intel community and the Department of Defense, towards more commercial oriented acquisition of weapon systems and platforms, how can industry play a hand in helping force this change, given a lot of times they’re locked out of classified conversations or improperly CUI classified things? Thank you.

ANGELSON: All good. Thank you very much. Dr. Goodman’s question first.

This is not a uniquely American problem. Because of the vast array of agencies—eighteen elements in the intelligence community—ours is probably more dramatic and more of immediate aid—immediate need of attention. But yeah, I mean, the Brits are—you know, and the Australians and the other members of the Five Eyes and the second five are looking at the same stuff that we are.

COHEN: Can I just make one—

ANGELSON: You may.

COHEN: —one asterisk on that very quick? We are the only country, though, where there is a presumption that every single thing that is done by the government will ultimately be reviewable by the American people, whether it’s in twenty-five years or seventy-five years. That is unique.

BURWELL: And for good reason.

SIMS: Yes.

ANGELSON: And last—so, and a ten-second answer here: What can—what can industry do to help?

COHEN: Michael, I think that’s—you’re in charge of the industrial.

THOMAS: Also, we’re the National Industrial Security Program. I got a lot of hats.

I think, just to tie it up very quickly, we at ISOO are the nexus between the defense industrial base and sort of the classified universe of information protection holders. So we do hear concerns about CUI and we do work regularly on reform there and I’d be happy to talk to you more about it offline, yeah.

ANGELSON: Thank you, everybody, for coming. God bless. (Applause.)

(END)

This is an uncorrected transcript.