U.S. intelligence agencies prohibit their former employees from writing or speaking about their government service without first obtaining government approval. The system, established after the Second World War to prevent the inadvertent disclosure of national security secrets, was initially modest in scope, but over a period of decades it has metastasized. Today, every intelligence agency imposes a lifetime prepublication review requirement on at least some of its former employees. Many agencies impose prepublication review obligations without regard to whether the covered employees ever had access to sensitive information, and without regard to how long ago those employees left government service. Submission requirements and review standards are vague, confusing, and overbroad. In the absence of concrete deadlines, manuscript review frequently takes weeks or even months, which means that books, articles, and blog posts cleared for publication are published long after the debates they seek to engage have subsided. Agencies’ censorial decisions are often arbitrary, unexplained, unrelated to national security concerns, or influenced by authors’ viewpoints.
The dysfunction of the prepublication review system has far-reaching effects, as the dispute overformer National Security Advisor John Bolton’s book highlighted—and as the Knight Institute and ACLU argue in Edgar v. Haines, a petition for certiorari now pending before the Supreme Court. Most significantly, the system impoverishes public discourse about foreign policy, national security, and war. Former intelligence agency employees often have unique insight into the operations and policies of the agencies for which they worked. Because their manuscripts draw on their personal knowledge and experience of government, their voices are not fungible or replaceable. The prepublication review system, at least in its current form, deprives the public of information and insight that is important to its ability to understand government policy, advocate for change, and hold government officials accountable for their decisions.
The system is in dire need of reform, but what would reform look like? Professors Jack Goldsmith and Oona Hathaway sketched some ideas in a series of blog posts and op-edspublished several years ago. Perhaps inspired by their suggestions, the House and Senate intelligence committees instructed the director of national intelligence in May 2017 to prepare a new prepublication review policy by October 2018 that would apply to all intelligence agencies and “yield timely, reasoned, and impartial decisions that are subject to appeal.” The new policy, the intelligence committees said, should require each intelligence agency to develop and maintain its own prepublication review policy identifying the individuals whose work is subject to prepublication review; providing guidance on the types of information that must be submitted for review; mandating timely responses; establishing a “prompt and transparent” appeals process; supplying guidelines for the assertion of “interagency equities”; and summarizing the measures agencies may take to enforce their policies.
All of this was promising. Implementation of the intelligence committees’ instructions might not have addressed all of the prepublication review system’s problems, but it would have addressed many of them. Three years after the committees’ deadline, however, no new policy has been produced, and there is no evidence that one is forthcoming. Perhaps there is more going on behind the scenes, but from all appearances the Office of the Director of National Intelligence (ODNI) seems simply to have disregarded the committees’ instructions.
The new administration, and the new Congress, should act more decisively to reform this broken system. As part of a broader effort to strengthen transparency and accountability in the national security sphere, President Joe Biden should issue an executive order that clarifies and narrows submission and review criteria, establishes new procedural safeguards, and mandates transparency about the prepublication review system’s operation. To ensure that future administrations do not backtrack on these reforms, and to impose additional safeguards against arbitrary and politicized decision-making, Congress should reinforce the executive order with legislation. Between them, the executive order and legislation would ensure that the prepublication review system strikes a more defensible balance between the government’s legitimate national security interests, the First Amendment interests of would-be authors, and the First Amendment and larger democratic interest of the public in ensuring that public debate about national security policy is fully informed.
Below, building on the intelligence committees’ instructions, we propose 10 guideposts for reform.
STREAMLINE THE SYSTEM
FIRST, the president and Congress should narrow the universe of former employees on whom the intelligence agencies can impose submission obligations.
One major problem with the current system is that it subjects far too many people to prepublication review obligations. When the system of prepublication review was first established, it applied to only a relatively small number of intelligence agency employees who had had access to the government’s most closely held secrets. Even when President Ronald Reagan briefly extended the system across the nation’s intelligence agencies, he extended it only to employees who had had access to sensitive compartmented information (SCI)—at the time “a very small fraction of Government employees who [had] access to classified information generally,” as former Deputy Assistant Attorney General Richard Willard testified to Congress. (The General Accounting Office reported that 119,000 agency employees had SCI access at the time.) Today, however, many agencies impose lifetime prepublication review obligations even on employees who have never had access to SCI, or even (in some cases) to classified information of any kind. The result is that millions of former government employees—more than 1.5 percent of the U.S. population, according to one estimate—are now subject to some sort of prepublication review requirement.
The president and Congress should limit the universe of former employees on whom the intelligence agencies may impose prepublication review requirements. Specifically, they should bar agencies from imposing prepublication review requirements on former employees who (i) have not held top-secret (TS)/SCI clearance, or (ii) who left government service more than 10 years ago.
Limiting the system in these ways would confine prepublication review to the circumstances in which it is most justifiable. It is important to remember that prepublication review has never been an effective safeguard against intentional disclosures of classified information, because those who disclose classified information intentionally do not submit their manuscripts for review. Instead, the narrower interest served by prepublication review is in preventing inadvertent disclosures. That interest is served most directly not by prepublication review, but by threat of administrative, civil, and criminal sanctions for the unauthorized disclosure of classified information. The interests served by prepublication review are narrower still with respect to the two categories of individuals identified above—those who have never held TS/SCI clearance and those who left government service more than a decade ago—because these two categories of individuals are unlikely to possess information whose inadvertent disclosure could cause serious harm. Any residual interest the government has in nonetheless reviewing their draft publications could be served by a system of voluntary submission.
Narrowing the universe of people subject to mandatory prepublication review requirements would have national security benefits too. It would allow reviewers to focus their attention on the manuscripts most likely to contain information whose disclosure would cause real harm. It would also relieve an overburdened system, making prepublication review faster and thereby reducing the likelihood that former employees disregard their obligations.
SECOND, the president and Congress should limit the kinds of materials agencies can require former employees to submit for review.
In the current system, submission requirements vary considerably by agency, and they are imposed through a confusing and sometimes conflicting tangle of contracts, regulations, and policies. As a rule, agency submission requirements are vague and overbroad and fail to give former employees fair notice of what they must submit. Many of them use phrases like “relates to,” “pertains to,” and “might be based upon,” which are of uncertain meaning and scope, and which invest reviewers with broad discretion that can easily be abused.
The CIA’s standard secrecy agreement, for example, requires former agency employees to submit all materials that are “intelligence related.” The regime of the Department of Homeland Security (DHS) obligates former DHS employees to submit all manuscripts “that reference DHS intelligence data or related activities, at any classification level, or … information derived as a result of affiliation with DHS.”Form 4414, the standard form that agency employees must sign in order to be afforded access to SCI, requires the submission of any materials that “contain or purport … to contain any SCI” or that “produce or relate to SCI or that [the individual has] reason to believe are derived from SCI.” Many of the operative terms here are sweeping and elastic. Former agency employees who have asked agencies for guidance about their submission obligations have been given inconsistent responses, when they have been given responses at all.
The president and Congress should limit the materials subject to prepublication review to manuscripts reasonably likely to contain or be derived from classified information obtained during the course of an individual’s government service. Limiting submission requirements in this way would clarify and narrow former employees’ obligations, align the prepublication review system more closely with the First Amendment, and allow reviewers to focus on the relatively small number of manuscripts likely to contain information whose disclosure would cause real harm.As the House and Senate intelligence committees have recognized, it would also “better incentivize” former employees to comply with their obligations.
ESTABLISH NEW PROCEDURAL SAFEGUARDS
THIRD, the president and Congress should strictly limit the time agencies may take to review the manuscripts submitted to them.
In the current system, former employees who submit manuscripts for review routinely wait weeks or months for a substantive response. The CIA now estimates that its review of book-length manuscripts will take more than a year. Some authors have waited even longer.The long delays deter some former employees from putting pen to paper. When they do write, their manuscripts sometimes do not see the light of day until long after the public debates they seek to engage have subsided. Frequently, the public is denied timely access to information and insight that is important to its ability to understand government policy or hold government decision-makers accountable for their decisions.
The president and Congress should impose clear deadlines for the completion of manuscript review. As an initial matter, they should establish a schedule of deadlines for review tied primarily to the length of the submission. One reasonable approach might be to require agencies to review manuscripts of less than 1,500 words within three days; manuscripts of 1,500–10,000 words within 14 days; and manuscripts of more than 10,000 words within 30 days. The president and Congress should also establish a channel for expedited review of newsworthy and other works whose publication is time-sensitive—an analog to the “expedited processing” provision of the Freedom of Information Act.
The president and Congress should also establish clear deadlines for interagency referrals. They should require agencies to make referrals promptly (e.g., within 24 hours for short manuscripts), to notify authors of the agencies to which their manuscripts were referred, and to notify them of the dates on which the referrals were made. They should also strictly limit the time an agency may take to review a manuscript referred to it by another agency. An agency that receives a manuscript through referral should be required to review the referred manuscript within the same period of time it would have been afforded had the manuscript been submitted to it in the first instance.
The president and Congress should also incentivize agencies to meet these deadlines. They should do this in two ways. They should provide that an agency’s failure to complete a manuscript review within the statutorily prescribed period will preclude the agency from imposing any direct or indirect sanction on the submitter for failing to satisfy her prepublication review obligations. (Direct sanctions would include attempting to impose a constructive trust on the submitter’s book proceeds. Indirect sanctions would include revoking the submitter’s security clearance.)
Finally, Congress should establish a cause of action that (i) permits a former employee whose manuscript has not been reviewed within the statutorily prescribed deadline to obtain an injunction requiring the agency to complete the review within a period of seven days; and (ii) entitles the former employee to reasonable attorneys’ fees and costs associated with the action.
FOURTH, the president and Congress should prohibit agencies from exploiting the prepublication review process for ends unrelated to national security.
Prepublication review obligations are quintessential prior restraints. The argument for their constitutionality turns on the claim that the restraints are necessary to protect national security secrets. Many agencies, however, use the prepublication review process to control and suppress other types of information. Indeed, some agencies expressly claim the authority to censor information unrelated to national security. For example, the DOD permits reviewers to censor any information “requiring protection in the interest of national security or other legitimate governmental interest.” Other agencies, including the National Security Agency and ODNI, fail to specify any limitation on reviewers’ censorship authority at all, effectively giving reviewers a free hand. Even those agencies whose regulatory frameworks seem to contemplate that reviewers’ authority will extend only to national security secrets permit reviewers to censor classified information without regard to whether the submitter learned of the information as a result of her government service. Agencies routinely censor information that submitters learned from newspapers, congressional hearings,and even the agencies’ own websites.
The president and Congress should prohibit agencies from using the prepublication review system to censor anything other than properly classified information that former agency employees learned in the course of government service.
FIFTH, the president and Congress should require agencies to document the reasons for their censorial decisions.
Currently, agencies are not required to provide authors with reasons for their decisions. The president and Congress should require agencies to provide authors with reasons, to the extent that reasons can be provided in unclassified form, and to document their reasons more fully in classified form. They should require agencies to make the classified record available to any court called on to review the propriety of the agency’s decisions.
SIXTH, the president and Congress should require agencies to establish an effective administrative appeals process.
Although some agencies currently have appeals processes, they provide no specific timeframes for resolution of appeals. For example, the ODNI states that appeals will be adjudicated “as time and resources allow.” Similarly, the Defense Department informs submitters that appeals will be resolved “as quickly as possible.” In practice, the appeals process is often a source of significant additional delay. Yet a “prompt and transparent” appeals process is essential, as the intelligence committees have recognized. The president and Congress should require agencies to resolve administrative appeals within a period of time equal to the statutorily prescribed time for initial review.
SEVENTH, the president and Congress should provide for prompt and meaningful judicial review of the agencies’ substantive decisions.
In the current system, agencies effectively have the last word with respect to what gets published. Submitters rarely challenge agency decisions in court because litigation is costly and time-consuming, and because courts almost always defer to agencies’ substantive decisions without assessing whether publication would cause harm or whether the risk of harm outweighs the public interest in disclosure.
The president and Congress could mitigate the first of these problems by establishing a “reciprocal notice” framework, under which an agency would be obligated to initiate judicial review upon notice from an author that she intends to contest the agency’s substantive decisions. The 2015 USA Freedom Act provides a possible model. Under that statute, if the recipient of a “national security letter” notifies the government that it intends to contest the legitimacy of a nondisclosure order, the government is required to initiate judicial review within 30 days, and the district court is required to “rule expeditiously.” In adapting this model to prepublication review, the president and Congress should set out short, mandatory deadlines for the government to file suit, and Congress should provide deadlines for judicial resolution of authors’ claims. Particularly in cases involving time-sensitive publications, it would be reasonable to require the government to initiate judicial review within five days of receiving an author’s notice, and to require courts to resolve the suits at least as quickly as they would resolve requests for a preliminary injunction.
To address the second problem identified above, Congress should require that courts reviewing agencies’ substantive decisions independently balance the government’s interest in secrecy with the public’s interest in disclosure. This balancing of interests is constitutionally required. As the Supreme Court has made clear, the question of whether information is properly classified is distinct from the question of whether an injunction against its publication is consistent with the First Amendment.
EIGHTH, the president and Congress should require agencies to publish their prepublication review policies.
Under the current system, former government employees often don’t know what their obligations are because the obligations are imposed through contracts, regulations, or policies that they do not have and that aren’t easily available online. In some cases, these documents aren’t public at all. For example, one CIA regulationrelating to prepublication review was made public only after the ACLU and the Knight Institute sued for its release. (A version of the document was already available online, but that version was 11 years old and heavily redacted.)
The president and Congress should require every agency that imposes prepublication review obligations on former employees to publish the contracts, regulations, and policies that impose or interpret the obligations. Building on the intelligence committees’ instruction, agencies should also be required to publish a “summary of the lawful measures each agency may take to enforce its policy, to include civil and criminal referrals.” The president and Congress should require the ODNI to consolidate all of this information on a single website accessible to the public.
NINTH, the president and Congress should require the ODNI to audit agencies’ prepublication review practices and policies and to report regularly to Congress and the public on the operation of the prepublication review system.
In the current system, even basic statistical information about prepublication review is difficult to come by. If responses to FOIA requests filed by the ACLU and Knight Institute are any guide, it seems that many agencies do not have this information themselves. The president and Congress should require agencies to maintain databases tracking at least the following with respect to every submission: the name of the submitter, the title of the submission, the date of submission, the type of submission, the length of the submission, the agency that received the submission, the agencies to which the submission was referred, the date of any substantive response, and a summary of that response. They should require agencies to make their databases available to the ODNI on an ongoing basis, and they should require the ODNI to publish a consolidated database in a native database format, excluding only the name of each submitter and the title of each submission (which could be added to the database after manuscripts are published). They should also require the ODNI to periodically audit individual agencies’ compliance with new statutory requirements, and Congress should require the ODNI to report regularly to the intelligence committees and the public about deficiencies in agencies’ compliance with those requirements.
PROVIDE DEDICATED FUNDING
TENTH, Congress should give intelligence agencies the resources they need to administer a streamlined prepublication review system effectively.
Some of the problems associated with the current system stem from a lack of resources. Streamlining the system along the lines proposed above would help a great deal. But Congress should also provide the agencies with dedicated funding for prepublication review. This funding should be separate from the funding provided to the agencies for declassification review and the processing of Freedom of Information Act requests. A better functioning prepublication review system should not come at the cost of these other systems, which serve important functions in our democracy.
* * *
In its current form, the prepublication review system imposes intolerable costs on former public officials who want to responsibly share their experiences and insights with the public; on the intelligence agencies, whose energies and resources the current system misdirects; and on the public, which is denied timely access to important information and ideas. It’s clear what steps need to be taken to fix the system. The new administration and the new Congress should take those steps quickly.
Jameel Jaffer & Ramya Krishnan, Opinion, We May Never See John Bolton’s Book, N.Y Times, Jan. 30, 2020, at A27.
Michael S. Schmidt & Katie Benner, Justice Dept. Scuttles Inquiry and Lawsuit On Memoir by Bolton, N.Y Times, June 16, 2021, at A16.
Edgar v. Haines, 2 F.4th 298 (4th Cir. 2021).
Jack Goldsmith & Oona A. Hathaway, Opinion, The government’s prepublication review process is broken, Wash. Post (Dec. 25, 2015), https://www.washingtonpost.com/opinions/the-governments-prepublication-review-process-is-broken/2015/12/25/edd943a8-a349-11e5-b53d-972e2751f433_story.html [https://perma.cc/4EXZ-2VS4]; Oona Hathaway & Jack Goldsmith, More Problems With Prepublication Review, Just sec. (Dec. 28, 2015), https://www.justsecurity.org/28548/problems-prepublication-review/ [https://perma.cc/NC24-WXK5]; Oona Hathaway & Jack Goldsmith, Path Dependence and the Prepublication Review Process, Just Sec. (Dec. 28, 2015), https://www.justsecurity.org/28552/path-dependence-prepublication-review-process/ [https://perma.cc/VCP4-QVWS]; Oona Hathaway & Jack Goldsmith, The Scope of the Prepublication Review Problem, and What to Do About It, Just Sec. (Dec. 30, 2015), https://www.justsecurity.org/28585/scope-prepublication-review-problem/ [https://perma.cc/VF93-ZKQP].
H.R. Rep. No. 114–573, at 7 (2017) (Conf. Rep.).
Brief Pet’rs Amicus Curiae, Edgar v. Haines, 2 F.4th 298 (2021).
Department of Homeland Security, Management Directive No. 11043, Sensitive Compartmented Information Program Management (2004).
Office of the Directorate of National Intelligence, Form 4414 –Sensitive Compartmented Information Nondisclosure Agreement.
United States v. Marchetti, 466 F.2d 1309 (4th Cir. 1972).
Cent. Intel. Agency, Protecting Secrets: CIA’s Prepublication Review Process (2017).
Scott Shane, C.I.A. Demands Cuts in Book About 9/11 and Terror Fight, N.Y Times (Aug. 25, 2011), https://www.nytimes.com/2011/08/26/us/26agent.html [https://perma.cc/93V3-XZ2T].
Thomas Reed Willemain, A Personal Tale of Prepublication Review, Lawfare (Jan. 10, 2017, 2:00 PM), https://www.lawfareblog.com/personal-tale-prepublication-review [https://perma.cc/XBU8-G9ZX].
New York Times Co. v. United States, 403 U.S. 713, 91 S. Ct. 2140, 29 L. Ed. 2d 822 (1971).
Cent. Intel. Agency, AR13, Special Reporting Requirements/Restrictions (2011), https://s3.documentcloud.org/documents/5767103/AR-13-10-Agency-Prepublication-Review-of-Certain.pdf [https://perma.cc/L3AH-55CF].
Cent. Intel. Agency, AR 13, Special Reporting Requirements/Restrictions (2011), https://irp.fas.org/cia/prb2007.pdf [https://perma.cc/248L-TXNB].
Jameel Jaffer is executive director of the Knight Institute.
Alex Abdo is the litigation director of the Knight Institute.
Meenakshi Krishnan was a 2019-2020 legal fellow at the Knight Institute.
Ramya Krishnan is a staff attorney at the Knight Institute.