Information Security Oversight Office (ISOO)

ISOO Audit & Audit Reports

Audit of the Withdrawal of Records from Public Access at the National Archives and Records Administration for Classification Purposes

April 26, 2006



WHAT THE AUDIT FOUND

Under the provisions of Executive Order (E.O.) 12958, as amended, "Classified National Security Information" (the Order) and in response to a request from the Archivist of the United States as well as a group of concerned individuals and organizations, the Information Security Oversight Office (ISOO) performed an audit of all re-review efforts undertaken since 1995 by agencies in their belief that certain records at the National Archives and Records Administration (NARA) had not been properly reviewed for declassification, but had been made available to the public. The audit found a total of ten unrelated efforts to identify such records, which resulted in the withdrawal of at least 25,315 publicly available records; approximately 40 percent were withdrawn because the reviewing agency purported that its classified information had been designated unclassified without its permission and about 60 percent were identified by the reviewing agency for referral to another agency for declassification or other public disclosure review.

In reviewing a sample consisting of 1,353 of the withdrawn records, we concluded that 64 percent of the sampled records did, in fact, contain information that clearly met the standards for continued classification. Much of this information had been declassified in the early years of implementation of the current framework before agencies had in place all of the required procedures and training. Agency declassification guidance was, at times, misconstrued and agency declassification personnel did not always recognize information that needed to be reviewed by other agencies. While these problems have been largely addressed over the years, we have concluded that more needs to be done.

The audit also found that in attempting to recover records that still contained classified information, there were a significant number of instances when records that were clearly inappropriate for continued classification were withdrawn from public access. We concluded that 24 percent of the sampled records fell into this category, and an additional 12 percent were questionable. In one re-review effort, the Central Intelligence Agency (CIA) withdrew a considerable number of purely unclassified records in order to obfuscate the classified equity that the agency was intent on protecting. Included in the inappropriate category above, at least 12 percent of the records sampled had apparently been properly declassified, but were later improperly reclassified.

In addition to CIA, the Department of Energy (DOE), the Department of the Air Force (USAF), and the Federal Emergency and Management Agency (FEMA) have each conducted re-reviews at NARA. In addition, NARA itself initiated three re-review efforts. Depending upon the re-review effort, the sample of records withdrawn clearly met the standards for continued classification anywhere from 50 percent to 98 percent of the time.

Even when a withdrawn record met the standard for continued classification, in a number of instances we believe insufficient judgment was applied to the decision to withdraw the record from public access. In many of these instances, withdrawal did little to mitigate the potential damage to national security especially if the record had been published elsewhere. At times, withdrawal could actually serve to exacerbate the potential damage by drawing undue attention to the record. Furthermore, a significant number of records that were withdrawn had actually been created as unclassified documents but were subsequently classified by CIA at the time of re-review (often 50 years later) solely because they contained the name of a CIA official in the list of individuals provided a copy.

There were a number of contributing factors to the issues identified by this audit. Sufficient quality control and oversight by both the agencies and ISOO has been lacking, as has proper documentation for declassification decisions. In addition, NARA has, at times, acquiesced too readily to the re-review efforts or withdrawal decisions of agencies. Additionally, NARA has not had the necessary resources available to keep pace with agencies' re-review activity, let alone the overall declassification activity of the recent past which has resulted in the accumulation of hundreds of millions of previously classified pages which require processing by NARA. The most significant deficiency identified by this audit, however, was the absence of standards, including requisite levels of transparency, governing agency re-review activity at NARA. Absent these, NARA along with CIA and USAF resorted to ad hoc agreements that, in retrospect, all recognize should never have been classified in the first place.

As a result of this audit, the affected agencies have agreed to abide by interim guidance that includes provisions that require the public to be informed that records have been formally withdrawn from public access at NARA due to classification action as well as how many records are affected. Prior to official promulgation in regulation, this interim guidance will be fully coordinated, to include an opportunity for public comment. In addition, in response to many of the challenges highlighted by this audit, the principal agencies involved in conducting classification reviews of records accessioned into NARA have agreed, in principle, to create a pilot National Declassification Initiative, in order to more effectively integrate the work they are doing in this area. This initiative will address the policies, procedures, structure, and resources needed to create a more reliable Executive branch-wide declassification program.

Finally, in a personal message attached to this audit (attachment 1), the Director of ISOO has indicated that it would be wrong to look at the audit results solely in the context of declassification and reclassification. While the issues identified in these areas are significant, he has indicated that they are reflective of challenges confronting the classification system as a whole. In response to the findings of this audit, the Director is writing to all agency heads asking for their personal attention in a number of critical areas, to include facilitating classification challenges and routinely sampling current classified information in order to determine the validity of classification actions. In addition, ISOO will be initiating a number of training efforts in support of these objectives. Finally, agency heads will be requested to provide a status report within 120 days on the action taken with respect to these initiatives as well as with regard to the recommendations contained within this audit report. ISOO will report publicly on these actions.

 

BACKGROUND

This audit was originally initiated in early January 2006 based upon concerns surfaced in a letter from Mr. Matthew M. Aid to Dr. Michael J. Kurtz of NARA dated December 6, 2005. In a subsequent memorandum dated January 27, 2006, Mr. Aid provided 15 exemplars of historical records withdrawn from public access. Following receipt of that package and at the request of the Archivist of the United States, this audit was expanded beyond focusing solely on current re-review efforts and includes all re-reviews since 1995 (the initial onset of the Order) that had resulted in records at NARA being withdrawn from public access because they purportedly contained classified national security information. Subsequently, a group of concerned individuals and organizations, in a letter dated February 17, 2006, requested that ISOO undertake an audit and publicly report on its findings.

The findings of this audit should not obscure that fact that in the over ten years since implementation of E.O. 12958, agencies have declassified almost four times as much information as was declassified in the prior 15 years, totaling over one billion pages. CIA, for example, has put significant effort into its declassification activities and is the only agency to have placed millions of these pages into an automated searchable system available to researchers. NARA's Presidential Libraries, with support from CIA, took the initiative to lead and fund a program to coordinate review of presidential library records, which facilitated and accelerated the declassification review and release of this information. Furthermore, both DOE and USAF have used risk management survey techniques that can be used by other agencies in determining how to best focus declassification reviews in the future.

All of the agencies that have been included in this audit have been exemplary in their cooperation and assistance. The scope of this audit could not have been accomplished in such a brief period of time had it not been for their responsiveness.

OBJECTIVE

The objective of this audit was to:

  • Identify the number of records withdrawn from the "open shelves" at NARA over the past several years because they purportedly contained classified national security information
  • Identify the agencies that required the withdrawal action
  • Identify the authorization and justification for the withdrawal
  • Through a statistically significant sample, determine the appropriateness of the classification action
  • Examine the effectiveness of NARA's internal processes and procedures and recommend improvements where required

 

POLICY CONSIDERATIONS

The Order governs the classification, handling, and declassification of classified national security information. It was effective October 14, 1995 and was most recently amended on March 25, 2003. Some key policy considerations that are relevant to this audit include:

  • Classification, initial as well as continued, is more than an assertion on the part of an agency. Per section 3.1(a) of the Order: "Information shall be declassified as soon as it no longer meets the standards for classification under this order." An essential standard for continued classification is that the "unauthorized disclosure of the information reasonably could be expected to result in damage to the national security ' and the original classification authority is able to identify or describe the damage" (see section 1.1(a)(4)). For 25-year-old or older permanent records of historical value, additional authorization is required to continue classification (see section 3.3(d) of the Order). Finally, section 1.7 of the Order sets forth prohibitions and limitations with respect to the use of classification authority, to include reclassification.
  • There is no requirement in the Order, per se, to review a record prior to it being subject to declassification. Rather, from the onset of the Order, it was expected that agencies would employ risk management principles and engage in reviews of records only when necessary. Under such a construct, it was never envisioned that there would be zero defects in the declassification process; rather it was accepted that some records that might otherwise meet the standards for continued classification would nonetheless be subjected to declassification. With respect to declassification, potential damage to national security would oftentimes be diminished by the very age of the information (25 or more years). Risk would be further mitigated in that those record series that contained the most sensitive information (e.g., an intelligence agency's operational files) could be exempted from automatic declassification at twenty-five years and thus would not be subject to declassification without review at that time.
  • A fundamental concept to this and prior Executive orders was the distinction between declassification and disclosure of information with or without proper authority. The term reclassification specifically refers to the classification of information after it has been declassified and released to the public under proper authority. Classified information that has been designated as unclassified without proper authority remains classified. Any subsequent public disclosure is unauthorized and constitutes a compromise of classified information.
  • Prior to the 2003 amendment of the Order, information could not be reclassified after it had been declassified and released to the public under proper authority (see section 1.8(c) of original Order). This section was re-designated section 1.7(c) in the amended Order and subsequently permitted the reclassification of such information if it had been declassified and released to the public under proper authority, but only under the following limited circumstances:
    - the reclassification action is taken under the personal authority of the agency head or deputy agency head, who determines in writing that the reclassification of the information is necessary in the interest of the national security;
    - the information may be reasonably recovered; and
    - the reclassification action is reported promptly to the Director of ISOO.
    It should be noted that since the March 2003 amendment to the Order and prior to the onset of this audit, no agency had reported to ISOO any reclassification action under this provision.2
  • Proper authority for the declassification and disclosure of information is addressed in section 3.6 of the Order (designated as section 3.7 in the original Order) which states that, "in response to a request for information under the Freedom of Information Act, the Privacy Act of 1974, or the mandatory review provisions of this order, or pursuant to the automatic declassification or systematic declassification review provisions of this order ... when an agency receives any request for documents in its custody that contain information that was originally classified by another agency, or comes across such documents in the process of the automatic declassification or systematic declassification review provisions of this order, it shall refer copies of any request and the pertinent documents to the originating agency for processing." This concept was further codified in the 2003 amendment to the Order with the addition of section 3.3(h) which states: "Records containing information that originated with other agencies or the disclosure of which would affect the interests or activities of other agencies shall be referred for review to those agencies and the information of concern shall be subject to automatic declassification only by those agencies'."
  • Section 3.6 of the order goes on to recognize that the identification of one agency's classified equities in other agencies' records could itself be classified information by noting that an agency can notify a requestor of another agency's classified equity in a record "unless such association is itself classified under this order or its predecessors."
  • The consequence of designating information as unclassified and disclosing it without proper authority was made more explicit when "Classified National Security Information Directive No. 1" (32 CFR Part 2001) (the Directive) was revised in September 2003 to add a new section 2001.10(d) which states: "Classified information that has been declassified without proper authority remains classified. Administrative action shall be taken to restore markings and controls, as appropriate."
  • Section 6.2 of the Order states explicitly that: "'Restricted Data' and 'Formerly Restricted Data' shall be handled, protected, classified, downgraded, and declassified in conformity with the provisions of the Atomic Energy Act of 1954, as amended, and regulations issued under that Act."

 

RECORDS WITHDRAWN TO INCLUDE JUSTIFICATION AND AUTHORITY

Since the Order was first implemented in 1995, we estimate that at least 25,315 publicly available records have been identified by agencies (to include NARA) and withdrawn from public access by NARA for classification concerns.5 Approximately 40 percent were withdrawn because the reviewing agency purported that its classified information had been designated unclassified without its permission and about 60 percent was identified by the reviewing agency for referral to another agency for declassification or other public disclosure review. These records were withdrawn during the course of ten re-reviews of publicly available records, the first of which was initiated in 1999.6

The damage to national security that could be expected to arise as a result of these more than 25,000 records being available for public research varied greatly depending upon the re-review effort. In some cases, it is clearly demonstrable that our national security would be at increased risk if the information involved was compromised. In other instances, the classified information being removed very frequently amounted solely to the name of a senior official at an intelligence agency's headquarters from the 1950's, even though the individual and his contemporaries are most likely dead, there was absolutely no indication that the official ever served under a cover assignment, and the agency itself had previously released the official's affiliation and position in other declassified records. Finally, in one re-review, purely unclassified records were withdrawn in order to obfuscate the classified equities that the agency was intent on protecting.

For the most part, from the perspective of DOE, CIA and USAF, the records that were re-reviewed should never have been placed on the open shelves in the first place, as the agencies believed that they did not have the opportunity to review the records as required by the Order. From their perspective, they were the "victims" of security practices that compromised their classified equities and damaged national security. From FEMA's perspective, it's re-review was an outgrowth of NARA's "Records of Concern" program post-9/11. From NARA's perspective, their three re-reviews were a consequence of quality control efforts that identified instances of NARA staff inadvertently exceeding their delegated authority to declassify or otherwise release certain records originated by other agencies.

  1. DOE Re-review Pursuant to "Kyl-Lott Amendment" (1999-Present)

    Pursuant to the National Defense Authorization Act of 1999 (PL 105-261, Section 3161), DOE has been conducting a re-review of records at NARA in order to preclude the inadvertent release of Restricted Data (RD) and Formerly Restricted Data (FRD).7 DOE reports to Congress on a periodic basis the results of this re-review, detailing the number of pages identified as containing RD or FRD as well as the total number of pages of RD/FRD withdrawn. (See: http://www.ssa.doe.gov/sp50/reports.html)
    The withdrawal of RD and FRD information was not included in this audit since it is not under the purview of the Order and is covered by a separate reporting regime to Congress. However, during the course of its re-review of over 35 million pages to date, DOE identified Federal records that they believed likely contained classified national security information under the Order (not RD or FRD) that was apparently declassified and released without proper authority.8 Similarly, they also identified information restricted by statute or regulation that was apparently released to the public without proper authority. DOE tabbed all of these documents for NARA to review or to refer to the responsible agency.
    To date, out of the boxes re-reviewed by DOE and re-processed and returned to the open shelves by NARA, the audit identified approximately 1,591 records tabbed for referral. In that NARA has been unable to keep up with the DOE efforts, these tabbed records have yet to be referred to the appropriate agency or returned to public access. In addition, there are an estimated 513 additional boxes that have been re-reviewed by DOE and require reprocessing by NARA, to include referral of the identified records. NARA has examined these boxes and estimates that there are at least an additional 573 records that have been identified by DOE for referral. This would bring the total to an estimated 2,164 records that await NARA review or referral to other agencies. This total was included in boxes that DOE had identified at the beginning of their effort and which could not be served to the public pending review by DOE and processing by NARA.
    DOE's re-review has been conducted pursuant to statute and is intended to address the inadvertent release of RD and FRD that had been declassified without proper authority. ISOO was aware of this re-review but was not aware of the referral activity until this audit. This referral activity was not included in DOE's periodic reports to Congress. While the referral of non-RD/FRD was not specifically covered by the statute, DOE believed that they were duty bound to identify instances where they understood classified national security information was improperly declassified, thus placing national security at increased risk, or where other controlled information was improperly released. Similarly, NARA believed that they were duty bound not to reintroduce the identified records into the public domain until their review or referral for review by the appropriate agency was completed.
     
  2. Central Intelligence Agency (CIA) Re-review of Certain Department of State (State), Bureau of Intelligence and Research (INR) Records (2000-2001)

    In 1999, representatives of NARA's Office of General Counsel noted what they believed to be inappropriately declassified records in a number of INR records contained in certain specific Lot Files. These records had actually been reviewed and declassified prior to the current Order and had been available for research since 1995. Subsequent inquiry revealed that this collection, totaling 55 boxes9 of records, was replete with errors. Due to a significant breakdown in communication between an agency reviewer and NARA, some classified records that State and CIA intended for release were withheld while some classified records that they intended to be withheld were released.10
    Once the source and magnitude of the error was recognized, NARA segregated these 55 boxes from the open collection. First State and then CIA initiated what would amount to several separate re-reviews, the last being completed in 2001. The CIA re-reviews subsumed the initial State review. Each of the CIA re-reviews cast a subsequently wider net resulting in a greater number of records being identified for withdrawal, to include a relatively small number of perceived classified equities belonging to other agencies. Overall CIA identified 3,147 records11 for withdrawal. Included in this were a significant number of what proved to be unclassified records that CIA identified for withdrawal in order to obfuscate the classified equities they wanted to protect.
    It is important to note that in its initial re-review of this collection, at the urging of NARA and others, CIA identified only about half of the total number of records eventually withdrawn as being of concern. Equally noteworthy, when CIA did a classification review of the initial withdrawals, they proposed to classify in full less than 1 percent, classify in part 68 percent, and release in full 32 percent. They eventually withheld all of these records plus another equally large group of unclassified records for the total of 3,147 records withdrawn. CIA chose this course of action in lieu of another option considered, the withholding of the entire collection of 55 boxes.
    Subsequent to these re-reviews, in order to prioritize the use of limited resources, NARA would not reprocess these boxes and actually withdraw identified records until there was a specific access demand for the box. Whenever there was a researcher request for a particular box, NARA would commit to reprocess the box on an expedited basis and as soon as possible. This would entail withdrawing identified records and substituting a withdrawal notice. The notice would be annotated with the actual date the box was reprocessed, not when CIA did their re-review; thus resulting in withdrawal notices with dates of recent vintage. NARA only completed these actions for this collection in April 2006.
    CIA notified their Congressional oversight committees in June of 2000, along with the intended corrective action. ISOO was also notified and determined that because of the many mistakes, this was not reclassification. CIA undertook this re-review because they believed that due to significant process errors classified information was subjected to declassification without proper authority and released to the public.
    The deliberate withdrawal of unclassified records for classification purposes is problematic for a number of reasons, not the least of which is that NARA assumed that all of the records identified for withdrawal contained classified national security information. Thus, when an access demand, such as a Freedom of Information (FOIA) request, was made on any of the withdrawn records, to include the unclassified records that were withdrawn for reasons of obfuscation, it is not readily apparent how these requests were handled. As such, this is a matter still under review by ISOO.
    In addition to the above, as a result of the multiple re-reviews conducted on this collection by CIA, the provenance of the collection has been undermined, in that the original order of the documents in many of the boxes was lost.
  3. CIA (Various Collections) (2000-2003)

    Largely pursuant to the now declassified memorandum of agreement between NARA and CIA, there have been at least two additional re-review efforts that have occurred with respect to at least six series of records at NARA's College Park facility, plus five additional records withdrawn on an individual basis at the request of CIA. As a result of these efforts, it appears that at least 780 records were withdrawn from public access. This activity was identified late in the audit and thus remains the subject of ongoing ISOO review.
     
  4. Federal Emergency Management Agency (FEMA) Re-review at Eisenhower Presidential Library (2002)

    In 2002, as part of efforts r elated to the "Records of Concern" initiative at NARA, the staff at the Dwight D. Eisenhower Presidential Library identified a number of previously released records that might support terrorist activity. These included 134 records that had been declassified and released following referral for review by FEMA in 1994 under E.O. 12356, the predecessor to E.O. 12958.
    During their records of concern review, the Eisenhower Library staff prudently identified potential classified national security information in some of the opened records and notified FEMA, which subsequently conducted an onsite re-review of these records under the Order. FEMA determined that two of the 134 documents reviewed were classified in their entirety and that the remaining 132 documents were partially classified and could only be released in part.
    The action taken by FEMA constituted reclassification of records that had been declassified and released to the public under proper authority, which was not permitted at the time that the action was taken. This issue is the subject of an ongoing review by ISOO. Additionally, the Office of Security for the Department of Homeland Security has informed ISOO that it will be initiating its own internal inquiry to determine the circumstances behind the FEMA actions and the steps necessary to correct any deficiencies and otherwise ensure compliance with the Order.
     
  5. NARA Re-review at Kennedy Presidential Library (2002-2005)

    As part of a quality control effort in 2002, the staff at the John F. Kennedy Presidential Library identified errors with respect to systematic declassification reviews conducted by Library staff in the 1990's. Management at the Kennedy Library and the Office of Presidential Libraries (NL) were concerned that members of the staff may have misapplied declassification guidance delegated to NARA and had declassified information in excess of NARA's authority. Upon discovery, NL assembled a team of experienced NARA reviewers and in July 2002, quickly assessed nearly 200,000 pages of records. As a result of this re-review, 405 documents were identified as potentially having content that exceeded NARA's delegated declassification authority.
    Following the re-review, the Kennedy Library staff decided to review yet one more time the identified records in order to cull out what they perceived to be the least sensitive. As a result of this subsequent effort by the Kennedy staff, only about half of the identified records were actually withdrawn from public access although all 405 records were referred in October of 2002 for review by other agencies. These agencies have since determined that: 319 of the records were either unclassified or suitable for declassification in their entirety; 62 records contained some classified national security information (these records were returned to public access in sanitized form); and,18 records required classification in their entirety. Agency decisions on six records remain pending.
    Between 2002 and 2005, the Kennedy Library staff undertook a separate effort to identify other collections within their holdings that had not been reviewed for classification prior to being opened to public access. In a number of instances, ISOO was notified. The staff focused on eight collections that were of concern and identified 612 records requiring classification review by other agencies. Following referral and review, these agencies determined that: 156 records were either unclassified or suitable for declassification in their entirety; 88 records contained at least some classified national security information (these documents were returned to public access in sanitized form); and, 25 records were classified in their entirety. The decisions on the remaining 343 records are still pending.
    Additionally, in May 2003, USAF personnel who were onsite for an equity identification training session with the Kennedy Library staff, identified and requested the withdrawal of one record for classified national security concerns. As a subsequent result of this training, the Kennedy Library staff withdrew three additional records from the open shelves and referred them to USAF. Following referral and review, USAF determined that three of the four records contained at least some classified national security information (these documents were returned to public access in sanitized form). The decision on the remaining record is still pending.
    In total, approximately 816 records were withdrawn from public access at the Kennedy Library. Of these, 423 (52 percent) have been returned to the public access, in whole or in part and 43 (5 percent) have been withdrawn in their entirety. Decisions are still pending for 350 records (43 percent).
     
  6. NARA Re-review at Bush Presidential Library (2002)

    As part of a quality control effort in September 2002, in response to an inadvertent release, staff at the George H. W. Bush Presidential Library identified a problem with the prior review and opening of a relatively small portion of the Library's holdings. Management at the Bush Library and NL were concerned that declassification guidance delegated to NARA had been improperly applied in some cases and that some information had been declassified in excess of NARA's authority. After reviewing the records at issue that had been previously released, Bush Library staff identified 318 records for referral to other agencies for re-review. These records included 97 records that were marked as containing classified national security information and 221 records not marked but suspected of containing classified national security information.
    Prior to the onset of this audit, these records had not yet been referred for declassification determinations. However, NL has recently begun an internal review of these records and the Bush Library staff will refer to other agencies those records that they believe they do not have the authority to declassify.
     
  7. USAF Re-review (2002-Present)

    In 1998, the USAF and CIA became aware that records revealing sensitive sources and methods of intelligence collection had been declassified without proper authority and released to the public. This came about, in part due to the failure of reviewers from various agencies to recognize the classified nature of the intelligence sources and methods involved.
    In 2002, USAF and NARA12 entered into a Memorandum of Understanding (MOU) that was classified SECRET and which applied to reviews conducted at NARA's College Park facility. This MOU was classified in order to protect the classified sources and methods at risk. The MOU went on to spell out how USAF would survey specific record series that were likely to contain previously declassified information related to the intelligence sources and methods of concern. If USAF reviewers discovered information that they believed had been improperly declassified and released to the public, NARA would be notified and the box segregated for a more intensive review by USAF. Improperly declassified records were then tabbed by USAF reviewers and were to be withdrawn by NARA when reprocessing the box for return to the open shelves. If an access demand was made for a particular box while it was segregated for review, USAF was to complete their review with a minimum of delay.
    Since 2002, USAF has identified a total of 17,702 records for withdrawal. Of these, 4,449 records were withdrawn in order to protect classified sources and methods at risk; an additional 2,263 documents were identified for withdrawal and referral to DOE; and approximately 10,990 records were identified for withdrawal for referral to more than 20 other entities, to include 1,444 for other USAF interests or activities. However, neither NARA nor USAF have been able to identify the number in the referral category with precision. Also, neither NARA nor USAF could readily identify which of these additional records were identified by USAF for referral due to reasons of classification and which were identified by USAF for referral due to other statutory and regulatory concerns. This matter remains subject to ongoing ISOO review.
    Subsequent to these re-reviews, NARA reprocessed these boxes with limited resources. This would entail withdrawing identified records and substituting a withdrawal notice. The notice would be annotated with the actual date the box was reprocessed, not when USAF did their re-review. Whenever there was a researcher demand for a box that had not yet been processed, NARA would commit to reprocess the box on an expedited basis and as soon as possible. To date, approximately 419 boxes that have been reviewed by the USAF remain to be reprocessed by NARA.
    Congressional oversight committees were notified when this issue was first identified. ISOO was notified and has received periodic briefings on this activity. ISOO had intended to audit this re-review but had not initiated any action in this regard until January 2006.
    USAF and CIA initiated their re-review because they believed that classified information was subjected to declassification without proper authority. USAF has largely completed its re-review of publicly available records. However, 200 boxes of additional concern were identified after the onset of the audit and are awaiting the end of the moratorium initiated by the Archivist of the United States.
     
  8. CIA Re-review of an "Internet Collection" (2005)

    In 2005, CIA became aware of a researcher's collection of previously declassified documents that became available over the Internet. CIA re-review of this collection identified documents that they believed had been improperly declassified. Subsequently, CIA conducted a re-review of the specific series at NARA from which these documents had been obtained, examining approximately 50,000 pages. Based upon this re-review, CIA identified 254 records for withdrawal, all of which represented CIA purported classified equities. CIA completed its efforts in October of 2005. It is worth noting that NARA appears to have processed the withdrawn materials in a timely manner and in step with CIA, resulting in minimal interruption in the availability of the boxes in which the information of concern was located.
    During the audit it was determined that both CIA and NARA failed to notice that approximately 28 of the items identified by CIA for withdrawal (11 percent of the total) and withdrawn by NARA had apparently been previously reviewed and declassified pursuant to proper authority under the FOIA. As such, the subsequent withdrawal of these records for classification constituted a reclassification action. To date, CIA has not taken the required actions necessary to reclassify information under the Order. This situation remains under ISOO review.
    ISOO was not notified of the problem or of this particular re-review activity. CIA initiated this re-review in the belief that classified information was subjected to declassification without proper authority.
    An overview of the re-reviews of publicly available records at NARA is provided in table 1 on the following page.
Table 1
Withdrawal Effort Dates Records Withdrawn
DOE (Referrals Only) 1999 -Present 2,16413
CIA (INR Collection) 2000 -2001 3,147
CIA (Various Collections)
(Two reviews plus individual records)
2000 -2003 780
FEMA - Eisenhower 2002 134
NARA - Kennedy
(Two reviews plus individual records)
2002 - 2005 81614
NARA - Bush 2002 318
USAF 2002-Present 17,70215
CIA (Internet Collection) 2005 254
Total   25,315

 

 

 

APPROPRIATENESS OF CLASSIFICATION ACTIONS

  • An objective of this audit was to select a statistically valid sample of each of the identified re-review efforts that resulted in the removal of previously declassified records from public access because they purportedly contained classified national security information.16 Using the standards for continued classification as set forth in the Order17 as well as available agency declassification guides, each sampled record was then assigned to one of the following three categories:
     
  • Appropriate - A record was considered appropriate if at least some of the information in the record clearly met the standards for continued classification.
     
  • Questionable - A record was considered questionable if the available declassification guide lacked specificity, if the audit team identified inconsistencies in how similar information was handled by agency reviewers elsewhere in the collection, or if the audit team, based upon experience, believed additional research was required to substantiate continued classification.
     
  • Inappropriate - A record was considered inappropriate if it was clear that none of the information in the record met the standards for continued classification or if the record was reclassified without following proper procedures.

One of the challenges confronted by this audit in determining the appropriateness of the classification actions taken by an agency was that it was not always apparent that the record was, in fact, initially declassified without proper authority. However, this is a challenge that similarly confronted each agency at the time of their re-review and each agency apparently took this into account with various degrees of consistency when identifying documents for withdrawal. In a number of instances, it appears that an agency simply assumed that the declassification had been done without proper authority. In fact, this audit identified a number of instances where the withdrawn document had been previously declassified under proper authority, oftentimes by the same agency that subsequently identified the document for withdrawal. This would occur, for example, in instances where the record had been previously released pursuant to a FOIA request. This could also occur when the record had been declassified by one agency using delegated declassification authority from another agency

The sampling results for the various re-reviews of publicly available records at NARA are as follows:

  1. DOE Re-review (Referrals Only)

    DOE has indicated that their referrals were not withdrawals per se and were made for additional reasons beyond classification, to include privacy concerns. As such, while ISOO did not sample these removals for purposes of this audit, this activity remains the subject of ongoing ISOO oversight.
     
  2. CIA Re-review (INR Collection)

    Based upon a sample of 427 records out of 3,147 records withdrawn, the audit revealed that 50 percent of the records sampled were appropriate for continued classification under the Order and the current CIA declassification guide as approved by the Interagency Classification Appeals Panel (ISCAP). However, a significant number of these records were classified solely because they contain the name of a CIA official in the list of individuals and offices that had been provided a copy of an otherwise unclassified document.
    The audit also revealed that 18 percent of the records sampled were questionable with respect to continued classification. These include the minutes and agenda of various intelligence board meetings where the members discussed the details of National Intelligence Estimates (NIE) for which the current classification status is unknown. Based on the age of the documents (late 1940s to early 1950s), the audit team believes it is likely that these records would fall into the inappropriate category with minimal research.
    Finally, the audit revealed that 32 percent of the records sampled were clearly inappropriate for continued classification under the Order. The unclassified records that were withdrawn for the purpose of obfuscation fell into this and the questionable category. In many cases, documents had been withdrawn for even the most innocuous mention of CIA. In many cases, records were withdrawn simply because the word "intelligence" was mentioned with no reference to CIA at all. CIA has acknowledged that they withdrew entirely unclassified records in order to obfuscate the classified equities they intended to protect in other records.
     
  3. CIA (Various Collections)

    As noted above, these withdrawals were only recently identified and were not sampled by ISOO. The activity remains the subject of ongoing ISOO oversight.
     
  4. FEMA Re-review (Eisenhower)

    As FEMA had originally declassified their equity in each of the 134 documents, the action taken in 2002 to classify information in these records constituted formal reclassification. In that reclassification was prohibited at the time that FEMA took the action, 100 percent of the 134 classification actions have initially been determined to be inappropriate (even if they might have met the standards for classification).
     
  5. NARA Re-review (Kennedy)

    Based on a sample of 251 records sampled out of 816 records withdrawn, the audit revealed that that 98 percent of the records sampled were appropriate for continued classification under the Order. The audit also revealed that 2 percent of the records sampled were questionable as to continued classification under the Order.
     
  6. NARA Re-review (Bush)

    Given the time limitations of the audit and the relatively low-threshold standards for classification of information less than twenty-five years of age (see section 1.4 versus section 3.3(b) of the Order), ISOO did not sample the withdrawn materials. Rather, this issue remains subject to ongoing ISOO review.
     
  7. USAF Re-review

    Based upon a sample of 403 records out of at least 17,702 records withdrawn, the audit revealed that 74 percent of the records sampled were appropriate for continued classification under the Order and the current declassification guide used by this program, pending formal approval by ISCAP.
    The audit also revealed that 18 percent of the records sampled were questionable with respect to continued classification. This was often attributed to a lack of specificity in the guide or due to the borderline nature of the information within the document.
    Finally, the audit revealed that 8 percent of the records sampled were inappropriate for continued classification under the Order. In many cases, the information that precipitated the withdrawal action was declassified in the most recent version of the declassification guide. However, the declassification guide has been revised several times over the past several years, each time authorizing the declassification of more information. As with those records in the questionable category, the information may have been authorized for continued classification at the time it was originally identified for withdrawal.
    As is the case with other re-reviews, USAF classification actions were more likely to be appropriate when tied to the classified intelligence sources and methods which were the original focus of the re-review (i.e., the information covered by the MOU). They were also more likely to be appropriate when the decision affected CIA. However, classification actions were less likely to be appropriate when they dealt with equities of the other agencies, of which there were at least 20.
    USAF has indicated that after they had completed their re-review of records on the open shelves, they intended to go back and re-review all of the withdrawn records and make a final determination with respect to the appropriateness of classification. There is every indication that USAF has proceeded deliberately at each step of their re-review with respect to the information covered by the MOU. USAF had anticipated returning a certain percentage of the withdrawn records associated with the MOU. During the audit, USAF provided ISOO with both procedures and plans to begin an effort to again review the items they withdrew.
     
  8. CIA Re-review (Internet Collection)

    Based upon a sample of 138 records out of 254 records withdrawn, the audit revealed that 78 percent of the records withdrawn were appropriate for continued classification under the Order and the current CIA declassification guide as approved by the ISCAP. However, many of these records (about 45 percent) were classified solely because they contain the name of a CIA official in the list of individuals and offices that had been provided a copy of an otherwise unclassified document. In many of these cases, the record was created as an unclassified record and handled as such with the CIA official's name handwritten into the margin after the document was created, and with no effort at that time to annotate the document as classified. In nearly every instance, the record was withdrawn due to the presence of one of two names, one of which had clearly been previously declassified by CIA, although perhaps in another context.
    The audit also revealed that 9 percent of the records sampled were questionable with respect to continued classification. These include documents that quote earlier NIEs, for which time did not permit the audit team to determine the current classification status. Based on the age of the documents (late 1940s to early 1950s), the audit team believes it is likely that these records would fall into the inappropriate category with minimal research.
    Finally, the audit revealed that 13 percent of the records sampled were inappropriate for continued classification under the Order. In 2 percent of the cases, it was apparent that documents had been withdrawn for even the most innocuous mention of CIA. In the remaining 11 percent, it was apparent that the record was withdrawn even though apparently the CIA itself, in response to a FOIA request, had previously declassified it pursuant to proper authority. The audit team went beyond the sample and reviewed all 254 records and confirmed that 28 records withdrawn fell into this category.
     

 

FINDINGS

  1. Records at NARA containing classified national security information were inappropriately designated as unclassified and released to the public.

    Most of the agency re-review efforts identified by this audit were precipitated, at least in part, by the failure of agencies to properly recognize the classified equities of other agencies. Most of these failures occurred in the 1990's and earlier, especially in the first years of the current Executive order. Much has been accomplished since then in the interagency process to enhance equity recognition, to include the CIA-led interagency External Referral Working Group (ERWG). This CIA-led effort has made interagency coordination more efficient, and has improved training on equity recognition so that the kinds of problems that led to the activity covered by this audit is much less likely to occur in the future. Additional interagency training has occurred such as the ISOO/USAF sponsored equity recognition training in 2005, and extensive similar training provided by USAF in prior years.
    However, the failure to properly recognize other agencies' equities cuts both ways and remains an issue. This audit revealed that agencies are more likely to inappropriately identify a previously declassified record for withdrawal if the purported classification is due to another agencies' equity rather than its own.
    Finally the re-review effort that accounted for 70 percent of all withdrawn records, the USAF re-review, was precipitated primarily due to the classified nature of the sources and methods of concern which contributed to the failure of previous reviewers, including USAF reviewers, to properly recognize the issue in the first place.

    Recommended Action: Implementation of a National Declassification Initiative discussed elsewhere in this report, to include additional training in equity recognition should aid in addressing this issue.
     
  2. Previously declassified records at NARA were removed from the public access even though continued classification was not appropriate.

    This audit identified a significant number of withdrawal actions for classification purposes as inappropriate. Of the records sampled to date, 24 percent were clearly inappropriate and 12 percent were questionable. In most of the agency re-reviews examined by this audit, records were inappropriately removed from the public access at the National Archives. In one case (the CIA re-review of the INR collection), 32 percent of the sample of withdrawals for classification purposes clearly did not satisfy the standards for continued classification under the Order, while an additional 18 percent were questionable.18 Notably, and as an exception, for the two re-reviews at the Kennedy Library, none of the sampled withdrawals were deemed inappropriate.

    Recommended Action: Agencies and NARA must develop and implement a strategy that will result in the prompt return to public access those records withdrawn to date that are not appropriate for classification.
     
  3. Agencies reclassified records that had been previously declassified under proper authority in violation of the Order.

    When conducting re-review activity, agencies did not properly take into account whether a record identified for withdrawal had, in fact, been declassified under proper authority. As such, agencies were often not in a position to recognize when they were undertaking reclassification action as laid out in the Order. For example, the USAF withdrawal action encompassed some documents that were declassified pursuant to E.O. 12937.19 In another re-review, CIA action encompassed documents that CIA had previously declassified pursuant to the FOIA. Overall, this audit identified more than 162 instances (12 percent of the records sampled) of likely reclassification actions that violated the provisions of the Order, to include an outright ban on such action prior to 2003. With additional research, it is likely that more such instances will be identified.

    Recommended Action: Due to the difficulty in properly distinguishing between records declassified with or without proper authority, the standards discussed elsewhere in this report governing any future re-review of previously declassified records that have been available for research at the National Archives must apply regardless of whether the action constitutes reclassification or because the record was declassified without proper authority.
     
  4. In at least one re-review, unclassified records were deliberately removed from the public access at NARA.

    In at least one re-review conducted by the CIA, unclassified records were withdrawn in order to obfuscate the truly sensitive classified equities the agency was trying to protect. Some of these unclassified records were subsequently the subject of FOIA requests, most of which have been pending for up to four years. The manner in which these requests have been handled is still a matter under review, with emphasis on ensuring that any FOIA requests that were denied on the basis of a national security exemption as set forth in 5 U.S.C. 552b(1),20 did, in fact, contain classified national security information.

    Recommended Action: The complexity of the issues pertaining to this situation require ongoing ISOO oversight.
     
  5. Sufficient judgment is not always applied to decisions to withdraw previously declassified records from the public access for classification purposes.

    Based upon review of withdrawn records as well as interviews with concerned personnel, it is apparent that records were often withdrawn because the agency can continue classification of a specific record, with infrequent consideration as to whether classification should be continued. Clearly, whenever information has been improperly declassified and made available publicly, the information has been subjected to possible compromise. When this occurs, agencies are required to ascertain the degree of damage to national security and to conduct a classification review. Part of this review process should be a determination of the extent to which the information has been compromised and whether withdrawal would serve to significantly mitigate the damage.
    Equally important is consideration of whether withdrawal itself could actually exacerbate the damage to national security, by drawing undue attention to the record in question. Standing alone, 25 or 50-year-old historical information is just that, historical information. Such information will often have little or no overt nexus to today's national security interests, unless an agency overtly creates an association, for example, by identifying the record for withdrawal, thus exacerbating the potential damage to national security. This was a point raised by NARA during the re-review, but which was either ignored or disregarded by various agencies. NARA now recognizes that in some instances they acquiesced too readily to some of the re-reviews and withdrawal of some records. In other instances, where NARA resisted, especially at the Presidential Libraries, agencies would, at times, reconsider and withdraw their request.
    In addition to the above, in a number of instances, the information withdrawn by agencies has been declassified elsewhere to include in the Foreign Relations of the United States (FRUS) as well as on the CIA's CREST21 computer database which is publicly available in the library at NARA. This system and the FRUS remains replete with declassified records that deal with the same intelligence matters that are dealt with in some of the withdrawn records, which again highlights the need for informed judgment when recommending the withdrawal of previously declassified records from public access.

    Recommended Action: Implementation of the standardized procedures discussed elsewhere in this report must address this issue, to include the need to ensure that re-review and withdrawal actions are rare and that collaboration between agencies and NARA with respect to determining the appropriateness of such action in the first place always occurs with provisions for challenge and appeal.
     
  6. Sufficient quality control and oversight is not always devoted to agency reviews.

    Errors with respect to the initial inappropriate declassification of records were compounded by the inappropriate classification activity taken with respect to certain records following agency review; indicative of inadequate quality control by agencies conducting reviews.
    In addition to acquiescing too readily to re-review efforts, NARA lacked the resources to monitor agency withdrawal actions. Time and effort devoted to monitoring agency withdrawal action detracts from NARA efforts to process and make available new records. Nonetheless, NARA understanding of: (i) the length of time the record has been available to the public; (ii) the extent to which the box containing the record has been requested for research; and (iii) the extent to which the record and/or the information at issue has been referenced and/or published in government and non-government publications is essential to ensuring that informed decisions are made with respect to any withdrawal action.
    In addition, ISOO oversight of agency review and withdrawal activity has been lacking.

    Recommended Action: The reviews of agency declassification efforts initiated by ISOO in the past year must be enhanced to include focusing on agency quality control procedures. Furthermore, quality control steps must be specifically incorporated into any future re-review activity that could result in the withdrawal of records from public access for classification purposes. Such quality control measures need to be implemented by both the agency conducting the re-review and by NARA. In addition, ISOO must audit and report on all such future re-review activity.
     
  7. Sufficient documentation is often not maintained for declassified records.

    A key principle of the current Order is that specific action and authority is required to continue classification beyond 25 years for historical permanent records subsequent to the onset of automatic declassification on December 31, 2006. As such, in order to determine the classification status of such information it will often be necessary to determine the following information: i) when an agency reviewed a document; ii) the results of the review; iii) the other agencies to which the record was referred as well as when; and, iv) the results and dates of other agencies' actions. This audit highlighted that documentation is often inconsistent, insufficient, and often provides more questions than answers, thus raising questions as to records' true classification status.
    This lack of consistent documentation significantly hampered this audit.

    Recommended Action: In coordination with affected agencies, ISOO must develop within 60 days specific guidance with respect to essential documentation that must accompany declassification actions. Implementation of the National Declassification Initiative described elsewhere in this report, including uniform databases, should aid in addressing this issue.
     
  8. NARA resources and processes have not kept pace with agency re-review and declassification activity at NARA facilities.

    Over the past ten years, unprecedented classification and declassification decisions have been made for records at or destined for NARA. Since 1995, agencies have significantly added resources and processes to their declassification reviews capability, putting further demands on NARA's limited capabilities. During that time period, there has been no corresponding increase in resources available to or otherwise applied by NARA for this activity.
    This mismatch in agency and NARA capabilities is highlighted by the fact that agencies could identify records for withdrawal quicker than NARA could reprocess the boxes and return them to the open shelves. Subsequent to agency re-reviews, in order to prioritize the use of limited resources, at times NARA would not reprocess boxes and actually withdraw identified records until there was an actual access demand for the box.
    To date, more than 932 boxes that had been segregated pending agency re-review remain to be reprocessed by NARA to the open shelves. This is contrary to the essence of the National Archives, whose mission, in large part, is to allow the public to discover on their own our nation's historical records and the information they contain. This is difficult to do if boxes are not immediately available for perusal. It is not unusual for researchers pressed for time to forsake an opportunity to review a box if they have to make a special request and wait several days. Access delayed can equate to access denied.
    The mismatch in capabilities is not abating. There are currently an estimated 450 million pages of records at NARA at College Park that have been processed by agencies for declassification but are pending archival processing and, where appropriate, withdrawal prior to being made available for research. This includes approximately 80 million pages identified for referral to other agencies due to the presence of other agencies' classified equities. Furthermore, agency action in the coming months to restore to the open shelves the thousands of records inappropriately removed will create additional workload for NARA.
    An additional consequence of this mismatch in resources and capabilities has been the need for NARA to provide broad stack access to those agencies conducting classification and declassification reviews at NARA. Such access diminishes NARA's control over the records and presents additional challenges in ensuring that the necessary administrative controls are in place and are appropriate to the environment in which access occurs and the nature and volume of information involved.
    In addition to the above, in a number of instances, to include records accessioned into the National Archives at College Park as well as records at one of the Presidential Libraries, records were withdrawn from the open shelves and segregated, but no attempt was made to refer or otherwise review the record in order to ensure that it was, in fact, appropriate for
    continued classification. This is especially so in those instances where one agency identified

    purported classified equities of another agency. Additionally, while resource intensive, NARA could have done more to monitor agency withdrawal action.

    Recommended Action: NARA must reengineer its processes and procedures and explore all avenues to enhance its resources in this area in order to ensure that records are processed and made available to the public as soon as legally possible. Additionally, NARA must examine and otherwise ensure that appropriate oversight is provided of agency actions regarding accessioned records. Implementation of the National Declassification Initiative discussed elsewhere in this report, as well as the development and implementation of standardized procedures governing the re-review of previously declassified documents at the National Archives should assist in addressing this issue.
  9. Standards, including requisite levels of transparency, governing the re-review of previously declassified records that have been available for research at NARA have not been promulgated.

    Because of the many issues associated with the CIA re-review of INR records and the resulting damage to the collection's provenance, NARA sought to establish more effective procedures in the event that any future re-reviews of open records might occur. Along those lines, in 2001, NARA entered into a classified MOU with CIA relating to records among the holdings of the National Archives at College Park, which was recently declassified. Subsequently, in 2002, USAF and CIA requested that NARA sign a similar classified MOU, to address a particular problem of improper declassification that they had identified. The USAF MOU was modeled directly on the CIA MOU. NARA has indicated that their interest in signing both MOUs was to ensure that the records were properly handled, that the re-review of open records was done expeditiously in order to minimize the burden on researchers, and to minimize the number of records withdrawn. During the course of this audit, the Archivist of the United States gave clear direction that never again would the National Archives enter into such classified agreements.
    Notwithstanding the intent, these agreements, and any other instances of limited guidance, failed to preclude the issues identified by this audit. To that end, standardized procedures must be developed and implemented that ensure that withdrawal of records from the public access are rare, are conducted in collaboration with NARA, and occur in as transparent a fashion as possible. Appeals procedures need to be included to address those instances where the agency concerned and NARA do not agree as to the appropriateness of the action.
    Furthermore, it has been pointed out that public reports made under the Kyl-Lott re-review program offer a useful example of the type of information that would help agencies, the Congress, and the public to understand what happens when agencies conduct re-review efforts that result in the withdrawal from public access of previously declassified records. Such transparency has been lacking in the other agency re-review efforts.22
    Increased transparency would help ensure that such withdrawal actions would occur only when absolutely necessary in the national interest and could dispel perceptions that such efforts are attempts to conceal official embarrassment or to otherwise attempt to "rewrite history."
    Finally, in an abundance of caution intended to preclude the inadvertent disclosure of classified information, withdrawal notices inserted in place of withdrawn records have contained a dearth of information to the point it makes it exceedingly difficult for researchers to determine if they want to make an access demand through FOIA or the mandatory declassification review provisions of the Order.

    Recommended Action: The draft protocol found at attachment 2 has been coordinated with the agencies that have conducted re-reviews of previously declassified records. This protocol includes provisions that require the public to be informed each time records are withdrawn from public access due to classification as well as how many records are affected. It also requires that withdrawal notices identify to the greatest extent practicable and with precision the record withdrawn and the reason for its withdrawal. The affected agencies have agreed to abide by this protocol as interim guidance until such time as it is formally coordinated and promulgated as a change to "Classified National Security Information Directive No. 1" (32 CFR Part 2001). Prior to promulgation, it will be fully coordinated, to include an opportunity for public comment.
     
  10. The current referral process is inadequate.

    While attempts have been underway for some time to revise the referral process to address deficiencies related to the documentation of referrals, the proposals generated to date fail to truly address the problems. To that end, a more fully coordinated Executive branch-wide approach to declassification of records, a National Declassification Initiative that better integrates individual agency efforts, is more reliable in results, and is more efficient in process is required. As a minimum, this initiative should:
    • enhance agency understanding of each other's sensitive information;
    • provide additional training that develops the needed understanding;
    • establish centralized databases and other resources to facilitate sound declassification decisions and provide the required documentation;
    • provide for greater consistency in the level of review applied to records;
    • preclude redundancies in security reviews;
    • increase the interface between declassification reviews done under the Order and those for other requests for access to information such as requests under the FOIA;
    • establish centralized priorities to include input from the Public Interest Declassification Board;
    • achieve greater efficiency and economy with respect to the utilization of finite resources; and
    • improve oversight.
    There are a number of examples where a concerted, Executive branch-wide approach has worked in the past two decades, such as the Iran-Contra, POW/MIA, Chile-Pinochet, and Nazi War Crimes reviews. Furthermore, the USAF re-review effort can, in many regards, serve as a model with respect to the development of risk management survey techniques and the use of databases. Finally, the Public Interest Declassification Board can also provide needed advice and guidance to the performance of this initiative, consistent with its responsibilities.

    Recommended Action: In response to many of the challenges highlighted by this audit, the principal agencies involved in conducting classification reviews of records accessioned into the National Archives have agreed, in principle, to create a pilot National Declassification Initiative, with the objective of more effectively integrating the work they are doing in this area. This initiative is intended to address the policies, procedures, structure, and resources needed to create a more reliable Executive branch-wide declassification program. The details of this proposal are currently being further developed and promulgated and within 60 days.

 

REVIEW METHODOLOGY

CIA - INR Lot Files

Processing of State Department INR Lot Files was still ongoing by both NARA and CIA at the time these boxes were sampled for the audit. Initially, the audit team was advised that there were a total of approximately 1,600 "tabbed" items associated with this particular re-review. Based upon this information, in order to review a statistically relevant number of records, the audit team selected a sample size of 304 tabs from 55 boxes. The team decided to examine a minimum of 75 percent of the boxes (42 boxes), reviewing 10 randomly selected tabbed items from each box. This would result in a sample size of 420 tabs.

The above approach encountered a number of impediments. For example, many of the boxes selected were not available to the audit team because NARA staff had not yet completed the reprocessing of the boxes, to include withdrawal of the tabbed items. In addition, some boxes were in the Interagency Referral Center (IRC) going through the referral process. At the same time, the entirety of the project was simultaneously undergoing scanning by CIA. Also at the time, the staff of the Nazi War Crimes and Japanese Imperial Government Records Interagency Working Group (IWG) was also reviewing some of the boxes. Furthermore, the audit team also encountered difficulty in determining exactly what documents were withdrawn from the series during its initial review in 1995 and what documents were withdrawn during the re-review in 2000-2001. The audit team ascertained that both the initial review and re-review projects had both been given identical project and tab numbers by NARA, necessitating the comparison of all withdrawn items with the withdrawal notices in the open stack boxes.

Resolution of the above issues was exceedingly time consuming and forced the audit team to limit the number of boxes reviewed. As the team could only review certain boxes on certain days, they took a larger sample from some boxes than initially planned in view of the significant time constraints under which the audit operated. In the end, the audit team was able to review tabbed items from 27 rather than 42 boxes.

The audit team's decisions as to the appropriateness of agency re-review decisions were based primarily on the use of CIA's ISCAP-approved declassification guide. The guide itself is very broad in the categories of information that may be exempted and the team did not have access to the CIA declassification handbook used by its reviewers.

In the end, 427 tabs were selected for the sample.

Afterwards, NARA reported to ISOO that following their withdrawal actions, the number of tabs identified for withdrawal by CIA was actually 3,147 as NARA had broken down some tabs identified by CIA into additional tabs.

USAF

For the boxes that USAF re-reviewed and that NARA had reprocessed to the open shelves, the audit team was initially informed by NARA that 8,868 records had been withdrawn for purposes of classification. Based upon this information, in order to review a statistically valid number of records, the team selected a sample size of 310 records. The "tabbed" items to be reviewed were contained within 589 specific NARA numbered projects. The audit team selected withdrawn items from every third project number on the list furnished by NARA. The withdrawn items for each project were typically contained within a single box. From each box the team originally planned to randomly select and review two "tabbed" (i.e. withdrawn) records. This approach would result in a total sample of 392 records.

In view of the severe time constraints associated with this audit, the audit team was left with no choice but to select the sample only from those boxes that NARA had reprocessed since NARA initially had no idea as to the volume or location of the boxes reviewed by USAF that were awaiting NARA reprocessing. It should be noted that the above referenced 8,868 items did not include the items contained within the boxes then not yet processed by NARA. As such, it was expected that the actual number would likely be significantly higher.

The above approach encountered a number of impediments. For example, many of the boxes were not available because they were in the IRC undergoing additional processing for referrals. USAF had also pulled many of the boxes to do an additional review on the material. In addition, the boxes of withdrawn items were not segregated in the stacks and were interfiled with unrelated projects. It usually took longer to find a specific box than it did to perform the review of its contents.

In the face of these obstacles, the audit team compensated for missing boxes by reviewing the next relevant box on the shelf. While NARA staff offered to locate any specific box that the team required, in the interest of completing the audit in a timely manner, the audit team decided to press on with the boxes it was able to readily locate.

Subsequently, USAF provided the team with a list of tabbed items by record group and a list of referrals to other agencies. These two lists combined resulted in a total of 17,702 tabbed items for withdrawal or referral. The audit team then examined what had been sampled to see if the records were representative of the various record groups involved. Based upon this examination, the audit team further adjusted the sample by reviewing additional records from a number of significant record groups that had not been properly represented in the initial sample.

The audit team's decisions as to the appropriateness of agency re-review decisions were based primarily on the use of a very detailed draft declassification guide provided by USAF. It must be noted that this guide has been updated several times since the beginning of the USAF re-review effort. Many of the documents that the team determined were inappropriate for classification may have been appropriate in the earlier version of the declassification guide. Since the projects were not entered into a database until NARA had completed its processing, the review dates reflected when NARA reprocessed the material, not when USAF re-reviewed it. Additionally, for those withdrawn records that were not associated with the intelligence sources and methods covered by the MOU but were instead other agency equities, there was typically little or no documentation associated with the referral.

Along those same lines, the audit team did not have the benefit of declassification guides or review handbooks for many of the agencies to which USAF referred records. Thus, the team evaluated the appropriateness of referrals based upon the standards of the Order as well as their own experience and knowledge gained through serving as staff to the ISCAP.

In the end, 403 withdrawn records were selected for the sample.

 

CIA Internet Collection

Though the CIA originally reported that 5,000 pages had been withdrawn from the open shelves during this re-review project, the actual number proved to be 1,243 pages (254 documents). The original number had mistakenly included records that had never been declassified and put on the open shelves as well as many records that had been identified for CIA re-review by earlier open shelf re-reviews (including USAF and DOE re-reviews) and were included in the referral count for those re-reviews. The total number of records withdrawn as a result of this re-review eventually proved to be 254 tabbed items from 77 boxes.

The audit team determined that the minimum sample size would be 137 tabbed items. The withdrawn items were stored within 6 boxes and the team decided to review 23 randomly selected documents from each box for a total sample size of 138 tabbed items.

The audit team's decisions as to the appropriateness of agency re-review decisions were based primarily on the use of CIA's ISCAP-approved declassification guide. The guide itself is very broad in the categories of information that may be exempted and the team did not have access to the CIA declassification handbook used by its reviewers to implement the guide.

In the end, 138 tabs were selected for the sample.

Kennedy Library

The Kennedy Library staff provided the audit team with a list of the various projects undertaken to review and withdraw records from public access for the purposes of classification. This list identified 1,031 documents that had been identified as requiring referral to agencies for declassification review. Of these, ten were for RD/FRD and thus beyond the scope of this audit. Of the remaining, four were for USAF, and 1,017 were for referral to other agencies. However, as noted earlier in this report, the Kennedy Library staff had actually withdrawn only about 816 records that were subject to the Order. Based upon this information, the team selected a sample size of 250 records.

The audit team was provided access to all of the guidance that was available to the Kennedy Library staff at the time the decision was made to withdraw the records. The team also evaluated the appropriateness of referrals based upon the standards of the Order as well as their own experience and knowledge gained through serving as staff to the ISCAP.

In the end, 251 withdrawn records were selected for the sample.

 


Footnotes:

1. This audit did not focus on withdrawal of records due to reasons other than the record purportedly contained classified national security information. For example, elsewhere in this report is discussed DOE's withdrawal of records containing Restricted Data or Formerly Restricted Data. Similarly, this audit did not encompass NARA's "Records of Concern" effort in which NARA re-evaluated access to some previously open unclassified archival materials in order to reduce the risk of providing access to materials that might support terrorist activity.

2. One such instance was reported to ISOO subsequent to the onset of the audit and remains under review by ISOO.

3. The methodology section of this report addresses many of the limitations of this audit due, in part, to the length of time covered (over 10 years), the inconsistent and incomplete records maintained by the agencies and NARA to document their activity in this regard, the time it takes for NARA to reprocess the boxes and return to the open shelves, and the ongoing use of the boxes to include satisfying researchers' requests. Furthermore, one of the challenges encountered by this audit is that different agencies counted the records they identified for withdrawal differently. Some agencies counted pages removed, others counted documents, and others counted "tabs," which are paper strips wrapped around documents to designate records to be withdrawn. A single tab placed by an agency might relate to what is later broken down by NARA at the time of actual withdrawal into multiple documents. This audit settled on counting that which resulted in a withdrawal notice - or tabs - since this is the consequence that is most apparent to the public. For the purpose of the audit, one tab was assumed to equal one record.

4. For purposes of this audit, a record was considered "withdrawn" if the reviewing agency determined that the record required classification or if the reviewing agency determined that the record had to be referred to another agency for declassification review. In such instances, NARA substituted a withdrawal notice for the purported classified or referred record once they re-processed a box.

5. While the audit focused on withdrawal for classification purposes, it must be noted that numerous items identified for withdrawal from public access by agencies were flagged not for reasons of classification but rather pursuant to statute or regulation. Such information includes content related to personal privacy, unclassified technical data with military or space applications, control of arms exports, certain intelligence agency information such as names of employees, and law enforcement information. Appropriate warning notices or other markings were often not present, recognized, or honored by agencies that conducted initial declassification reviews. While not the subject of this audit, it is clear that this issue will be an ongoing problem for NARA and the agencies.

6. We cannot preclude the likelihood that, in addition to the identified re-reviews of publicly available records, there have been an unknown number of isolated instances whereby one or more documents were withdrawn from public access due to classification. For example, we are examining a decision by CIA to withdraw four documents that had been previously processed and released in sanitized form as part of the Remote Archives Capture (RAC) effort at the Lyndon B. Johnson Presidential Library. Similarly, during the audit period, Johnson Library staff discovered and removed two documents from public access that had never been officially subjected to declassification.

7. RD pertains to certain information dealing with the design, manufacture or use of nuclear weapons. FRD pertains to information removed from the Restricted Data category upon a joint determination by DOE and DOD that such information relates primarily to the military utilization of nuclear weapons and that such information can be adequately safeguarded as classified defense information. Neither RD nor FRD information is subject to the provisions of the Order, and are instead classified pursuant to the Atomic Energy Act of 1954.

8. Per DOE, some of the record collections contained marked classified documents and were "bulk declassified" by other agencies without review.

9. In some documentation, this effort has been tied to 56 boxes of State Department INR Lot Files. However, one of the 56 boxes was from another series of State Department records. The items withdrawn from this separate box are included elsewhere in this report under the heading "CIA (Various Collections)."

10. At varying points in the review process, the handwritten designations placed on tabs by agency representatives were problematic. At one point "D" was intended or interpreted to represent "deny," at other points it was intended or interpreted to mean, "declassified." Similarly, at some points the designation "R" was intended or interpreted to represent "release," at other points it was intended or interpreted to represent "refer" (i.e., do not release).

11. As noted previously, when an agency places tabs in boxes, what they see as one document covered by one tab, might be subsequently broken down by NARA as multiple tabs or documents as part of the withdrawal process. Because of this, tab or document counts may very well differ when examining how much was identified for withdrawal by an agency and how much was actually withdrawn by NARA.

12. The MOU was between USAF and NARA's Office of Records Services and did not cover the Presidential Libraries. However, USAF conducted visits at various Presidential Libraries primarily to provide training and equity recognition of USAF programs. In the course of these visits, USAF identified a single document for withdrawal.

13. DOE believes it is inappropriate to list the records identified for referral as "withdrawn" in that these records were already removed from public access under the DOE review and could have been replaced on the open shelves once reviewed by NARA and other agencies except that NARA decided to defer this until after all the reviews were completed. Since, from a practical point of view, members of the public were denied access to these records, they are included here.

14. 424 of these records have since been returned to public access, in whole or in part, by NARA following decisions by pertinent agencies.

15. As was the case in footnote 14, USAF and other agencies has subsequently indicated that an indeterminate number of records could be returned to public access; however, NARA has yet to reprocess all of these records.

16. A total of 1,353 records were sampled as part of this audit.

17. It is important to note that all of the information that is the subject of the audit, except the 318 records at the Bush Library, is more than twenty-five years old, as is the case with the majority of the holdings of NARA. As such, the information is subject to a higher standard for continued classification (refer to section 3.3(b) of the Order) than information that is less than twenty-five years old (refer to section 1.4 of the Order).

18. Of the 15 exemplars provided on January 27, 2006, 11 had actually been withdrawn from public access. Of these 11, ten were from the CIA re-review of the INR collection.

19. On November 10, 1994, the President issued an Executive order that declassified, in bulk, a selection of classified records within the National Archives. This unprecedented Order declassified approximately 44 million pages or 14 percent of the National Archives holdings of classified material at that time, including classified holdings through the end of World War II, and a similar quantity that dated post 1945 and into the 1970s.

20. This exemption protects from disclosure national security information concerning the national defense or foreign policy, provided that it has been properly classified in accordance with the substantive and procedural requirements of an executive order.

21. CREST is "CIA Records Search Tool." It should be noted that although not a part of this audit, the audit team did become aware that previously declassified records that had been posted to the CREST system have subsequently been removed. ISOO will be following up on this issue separately with CIA.

22. Even the DOE public reports failed to note the multitude of records that were identified by DOE for referral to other agencies. However, DOE points out that their Reports to Congress were for inadvertent releases of RD and FRD, not for classified national security information (NSI). In instances where an RD or FRD document required referral to another agency because of the likelihood of also containing NSI, DOE did make note of such in footnotes within these reports. [See reports Three, Four, and Five.] Therefore, from DOE's perspective, they believe it is clear that they have never hidden the fact of referrals and have in fact made note of such activities (albeit only as it overlapped within a RD/FRD document) in its Reports to Congress. More importantly, DOE tabbed referrals were not withdrawals from their perspective, but advisement for NARA to review or refer to another agency to review.

 

Top