A Review of the Department of Energy Classification: Policy and Practice (1995)

Classification policy and the classification guides apply to categories and types of information. Once the information that is to be classified has been generated and defined, the task of classifiers and declassifiers is to determine whether a particular document contains classified information. Building on the general principles discussed in the previous chapter, this chapter addresses the question of what information should be classified. The difficult issues of identifying specific documents containing the information of interest and having them declassified and disseminated are the subject of Chapter 5.

There are public pressures for openness in four main areas: (1) the effects of Department of Energy (DOE) activities on health, safety, and the environment; (2) the exploitation of classified technologies with potential commercial applications;¹ (3) the historical actions of DOE and its predecessor agencies (the Atomic Energy Commission and the Energy Research and Development Administration); and (4) nuclear weapons policy, dismantlement of surplus weapons, and management of the resulting materials. Significant advances have been made in declassifying information in the first two areas; some progress has been made in the third and fourth. All information related to health, safety, and the environment has been declassified in principle.² Similarly, declassification of information with significant potential for commercial application has repeatedly occurred since the early days of the Atomic Energy Commission. Examples include declassification of information relevant to development of civilian reactors, information on reprocessing, and recently, most information relating to inertial confinement fusion. Significant issues remain to be addressed, however, especially regarding nuclear weapons information.

A well-informed national policy debate concerning the military application of nuclear weapons, nuclear nonproliferation, dismantlement of surplus weapons, and disposition of the resulting materials serves the public interest. We believe that public participation in such policy deliberations is necessary to obtain the broad public acceptance that will be needed to allow revised policies to be implemented effectively.³ But effective participation will not be possible unless the necessary information is declassified and made available in a timely manner. The Committee agrees with the view of President Reagan's Blue Ribbon Task Group on Nuclear Weapons Program Management:

DOE has only taken initial steps toward the declassification of information that will help to inform the public debate about nuclear weapons policy. Both the Secretary of DOE and the Director of the Office of Declassification have acknowledged the need to declassify more such information, insofar as possible, and on December 7, 1993, and June 27,

1994, the Secretary announced the declassification of certain additional categories of information. Although progress has been significant, the fact remains that some items of information having important bearing on weapons policy debates remain classified. In view of the changed world situation, the Committee concludes that consideration of further declassification actions is warranted.

Some of the areas in which there has been limited disclosure and that should be reexamined with a view to further declassification include

1. Yields of nuclear tests carried out since 1962. Knowledge of such yields would be extremely unlikely to provide information to potential proliferators about weapons design, but release of the data would serve U.S. national interests by making possible the calibration of seismic detection networks, thereby paving the way to more extensive foreign collaboration in nuclear test detection.

2. Information on weapons stockpiles and stockpiles of special nuclear material. A potential proliferator would in no way be influenced by the precise knowledge of these figures, which are enormously in excess of any aspirations of a proliferator.

3. Actions of DOE and its predecessors that are of significant historical importance. More aggressive declassification would greatly aid historical reconstruction of crucial actions regarding nuclear weapons policy, research, development, testing, and production and would be of invaluable assistance to scholars and policy analysts seeking to learn the lessons of the Cold War.

While the declassification of documents containing recently classified information, at least on a large scale, is a substantial and time-consuming undertaking, it should be possible to release information of major interest in summary form. For example, a request for information about test yields is in fact a request for specific information, not for documents. Releasing such information would show that the Department is committed to ending the practice of blanket secrecy on such matters, substituting instead a practice of releasing information that enriches public debate, while continuing to withhold information that contains legitimate

national security secrets. If serious questions about the accuracy of the information were to arise, the Department could ask the Information Policy Advisory Board to review the classified source documents to verify the accuracy of the information.

Such an approach might satisfy some of the public demand for information without the massive effort that declassification of the inventory of files will require. This approach, however, will not obviate the practical and legal obligation to produce documents eventually. Historians, policy analysts, and others understandably will demand the opportunity to review actual files. Some requesters, even if given information, will ask to see the actual documents to be sure that the information supplied is accurate and complete. Thus, the approach of providing information cannot be a substitute for the effort of declassifying documents. Indeed, the Freedom of Information Act (FOIA) imposes obligations to produce documents, not just information. Nonetheless, it offers a way to begin the process and may reduce the number and scope of FOIA demands.

At their summit meeting in January 1994, Presidents Bill Clinton and Boris Yeltsin agreed on the goal of ensuring the "transparency and irreversibility" of the nuclear arms reduction process.⁵ In addition, the United States is currently engaged in negotiations with Russia to persuade it to strengthen controls over the management of special nuclear material, as well as over spent nuclear fuel from the commercial fuel cycle. This urgent problem was highlighted by the recent, highly publicized smuggling of special nuclear material from Russia. A separate National Academy of

Sciences report addressed these problems in considerable detail.⁶ One major recommendation of that report is to greatly increase the public availability of information both in Russia and the United States concerning the content, location, and management of stockpiles of strategic and other nuclear materials.

Congressional actions clearly support release of nuclear weapons-related information on a reciprocal basis with Russia. The 1992 Senate resolution approving ratification of START I included the "Biden Condition," which required the President to seek arrangements to monitor stockpiles of weapons and fissile materials in the United States and the former Soviet Union using "reciprocal inspections, data exchanges, and other cooperative measures."⁷ The Defense Authorization Act for FY1993 permitted declassification of stockpile information in the context of an agreement between the United States and Russia for release of such data.⁸ As a further step in this direction, the Defense Authorization Act for FY1995 allows a one-year period (until December 31, 1995) in which the United States might reach an agreement for exchange of classified nuclear weapons information with Russia without requiring the full congressional review process provided by the Atomic Energy Act (AEA) for such international agreements.⁹

The Secretary's Openness Initiative has gone a considerable way toward the goal of transparency, and Russia has at least initially accepted some proposals to provide greater openness. For example, in March 1994 DOE Secretary O'Leary and Russian Minister of Atomic Energy Viktor Mikhailov reached agreement on unprecedented inspections of fissile

materials from dismantled nuclear weapons.¹⁰ During their discussions in June 1994, Vice President Al Gore and Russian Prime Minister Viktor Chernomyrdin agreed to establish a joint working group to explore the confidential, reciprocal exchange of data on stockpiles of nuclear weapons and fissile materials. This was followed in September by a Clinton-Yeltsin commitment to exchange stockpile data by the end of 1994.¹¹

As of early 1995, that exchange has not taken place. Russian secrecy, particularly in the nuclear energy field, is deeply ingrained and is yielding slowly and reluctantly to pressures for change. On the U.S. side, the delay lies in part in the need to achieve interagency agreement, but also on the strictures of the classification system. Much of the stockpile data that would be exchanged with Russia is restricted data (RD) or formerly restricted data (FRD). The U.S. has tabled a draft Agreement for Cooperation, but until that is signed by both countries, the data exchange cannot take place.

Overcoming the reluctance in both countries to reveal information long held secret requires sustained, high-level attention. DOE should continue to pursue reciprocal exchanges of information with Russia. Specifically, the Department should explore arrangements in which each party to the exchange retains the right to allow or prevent the public release of the information that it is providing to the other party, so that disagreements about whether information should be publicly released do not obstruct mutually beneficial exchanges.

The Committee concurs with the Academy's report on management and disposition on excess plutonium weapons about the urgency of measures to improve control and accounting of Russian fissionable materials and to improve transparency about the nuclear stockpiles of both the U.S. and Russia. However, the focus on this important objective should not be allowed to delay the release of declassifiable information about American nuclear weapons that is needed to enable informed debate about policies appropriate to the new era. The pursuit of a

confidential exchange of RD or FRD with Russia should not serve as a justification for delay in declassifying and releasing U.S. data.

In 1981 Congress gave the Secretary of Energy authority, under section 148 of the AEA, to prohibit the unauthorized dissemination of a category of information designated unclassified controlled nuclear information or UCNI. The AEA defines UCNI to include information pertaining to

1. the design of production facilities or utilization facilities;

2. security measures concerning such facilities or nuclear material that is contained in such facilities or is in transit; or

3. the design, manufacture, or utilization of any atomic weapon or component if that information has previously been declassified or removed from the restricted data category.¹²

DOE can designate information in these areas as UCNI if it determines that "the unauthorized dissemination of such information could reasonably be expected to have a significant adverse effect on the health and safety of the public or the common defense and security by significantly increasing the likelihood of (A) illegal production of nuclear weapons or (B) theft, diversion, or sabotage of nuclear materials, equipment, or facilities."¹³ Although UCNI is unclassified, it can be shared only with U.S. citizens in specified categories having a need to know, and failure to abide by the restrictions is subject to stiff sanctions, including a substantial civil penalty.

UCNI is controversial.¹⁴ Critics see the category as vague, with inclusion criteria that are difficult to distinguish from the criteria for classification.¹⁵ They charge it is being used too widely despite the strictures in the AEA to minimize its use. UCNI may also be subject to abuse; for example, it offers an easy way to avoid release of information requested under the FOIA.¹⁶ Critics argue that it creates barriers to information flow without clearly providing much protection against proliferation, adds cost and bureaucracy, and, in general, diverts resources from protection of more important information.

On the other hand, supporters argue that it is needed to protect sensitive information that could not otherwise be classified for legal or operational reasons. They also claim that it allows wider dissemination of information of relatively low sensitivity than would be possible if the only alternatives were classification or complete declassification.¹⁷

The Committee has not received persuasive justification for the retention of UCNI. The concept of information control beyond classification runs counter to the objective of focusing resources on protecting truly sensitive information. Moreover, the application of UCNI over the years seems to have drifted some distance from the language of the AEA and DOE's implementing regulation.¹⁸ As discussed below, the circumstances and context within which UCNI is applied have changed

sharply since the amendment was passed and DOE regulations developed. The Committee recommends a thorough reexamination of the need for UCNI. The Congressional Commission on Protecting and Reducing Government Secrecy or other appropriate body should reassess UCNI in the context of a broader review of controls on unclassified information.

In 1981 the principal reason for creating UCNI was to control information about sensitive nuclear facilities that might be of use to terrorists. ¹⁹ This information includes layouts for DOE buildings showing where radioactive materials are stored and where critical operations are conducted and emergency response plans at such facilities. Many uncleared people, such as police or firemen, might need such information to perform their jobs, but DOE does not consider it feasible to classify the large volume of documents involved and to provide security clearances to all people whose jobs require access to them.²⁰ Placing such information in the UCNI category enables DOE to make the information available without the cost, administrative burden, and delay of security clearances and classification procedures. At the same time, the availability of the information can be limited because of the penalties that can be imposed for unauthorized dissemination and because the statutory language provides a clear exemption from FOIA requests.²¹

According to the statutory provision described above, UCNI controls can be applied to information about the design of production and utilization facilities (i.e., nuclear reactors). These controls do not clearly apply to facilities for manufacture, assembly, or storage of nuclear weapons or their components—the facilities of greatest current concern. However, the Department has interpreted section 148 broadly to include such facilities, on the grounds that Congress intended section 148 to cover information related to nuclear weapons. DOE is considering a recommendation that the AEA be amended to clarify that all nuclear weapons facilities are covered.²²

The more fundamental question is whether UCNI is needed at all to protect information related to facility security. None of the studies addressing UCNI that were provided to the Committee contained any quantitative analysis that supports the argument that it is impractical to control security-related information about facilities through classification as NSI.²³ The Committee recommends that DOE evaluate the costs and feasibility of treating sensitive information related to facility security either as a special category of national security information (NSI) or, alternatively, as unclassified information not subject to special controls.

More recently, the growing emphasis on proliferation and the control of information in proliferation-related areas such as plutonium processing, isotope separation technologies, and high explosive technology related to nuclear weapons has expanded the range of information considered UCNI.²⁴ Although the original focus of UCNI was protecting information relating to the physical security of sensitive facilities, its use has evolved to control a broad spectrum of information on dual-use technology—that is, technology that can support both military and civilian applications.

DOE's study of classification policy raises questions about its authority to extend UCNI controls to proliferation-related information. The AEA focuses on "illegal production of nuclear weapons." This phrase does not clearly apply to production of weapons by sovereign nations or even by subnational groups outside the United States.²⁵ Moreover, the definition has important areas of ambiguity; the reference to "the design of production facilities or utilization facilities" could be interpreted as applying only to information about existing facilities that could be useful to a terrorist in planning an attack, not to more abstract information that could help a proliferator construct and operate such facilities.²⁶If UCNI

is to continue to be used as the basis for controlling proliferation-related technical information, the critical interpretations that provide the basis for that use should be included in an updated UCNI regulation.

While Exec. Order No. 12,958 provides for reclassification of NSI that has been declassified, the ''traditional'' interpretation of the AEA by DOE's Office of General Counsel has been that once an area of RD has been declassified, even new information in that area cannot be classified as RD, FRD, or NSI unless there was a caveat in the original declassification decision that contemplated such an action.²⁷ Thus, once an area is declassified, subsequently developed information in that area is not eligible for classification, even if it is highly sensitive. UCNI is understood, however, to allow the recapture of information that has been declassified. This opportunity is offered as a justification for preserving the UCNI category.

In our view, UCNI should not be used for classification of information that is highly sensitive, since the level of protection that can be afforded by UCNI controls is much lower than for information classified RD or NSI. In apparent recognition of this, DOE proposes an amendment to the AEA to allow classification of new information in areas that have been previously declassified.²⁸ Before seeking an amendment to the AEA, however, DOE should carefully reexamine the "traditional interpretation" that the AEA precludes such action. If a reinterpretation is possible, the proposed regulation dealing with RD discussed below should include a section providing for classification of new information in broad areas of information that have been declassified, based on a clear demonstration that uncontrolled release of this new information would cause an "undue risk" to national security. If a more favorable interpretation is not reached,

DOE should seek an amendment to the AEA to provide the authority to reclassify. In any case, DOE should not rely on UCNI to protect truly sensitive information.

Finally, DOE states that UCNI provided gradations in the control system that allow broader but not unlimited dissemination of information. ²⁹ DOE argues that an "all-or-nothing" classification scheme denies the opportunity to make the most appropriate use of information and that the change would be at least a partial move in the direction of the Department's Openness Initiative. It asserts that UCNI is used to serve a valuable purpose by allowing information of a lower level of sensitivity to be shared with the private sector for purposes such as commercialization, while still affording a degree of protection against nuclear proliferation.³⁰

UCNI seems poorly suited to this purpose. DOE's order implementing UCNI prohibits its use in place of classification controls under the AEA or executive order if classification would be appropriate and prohibits information from going directly from the category of RD to UCNI.³¹ Under these restrictions, the only way for nuclear information that had been classified as RD to be controlled under UCNI is for the information to be declassified and then for circumstances to change such that the information becomes sufficiently sensitive to warrant reassertion of some degree of control. Consequently, the current UCNI provisions do not appear to be particularly well suited to the objective of providing a "middle ground" in the classification spectrum.

An amendment to the AEA, or perhaps to the DOE order, might remedy this limitation, but the need to add intermediate levels of information control should be examined carefully before they are adopted. The creation of new categories of control is contrary to the basic premise of an effective system—tight controls over narrow areas.

If DOE concludes that information now encompassed by UCNI should continue to be protected under this scheme, it should prepare a clear and thorough background information document describing and explaining the rationale for the proposed uses of UCNI and a comparison of alternative approaches to achieving the same objectives. Currently available documentation is inadequate for that purpose. It is difficult, if not impossible, for an outsider, even with some diligence, to obtain a clear picture of the scope, application, and rationale for UCNI from available documents. The legally required quarterly reports on UCNI decisions provide little insight into the underlying rationale for the decisions. Furthermore, the two DOE classification policy studies provided to the Committee do not give an adequate discussion of options to allow the uninitiated to form an opinion.³² DOE cannot expect to be granted the benefit of the doubt by skeptics when it is difficult to obtain a clear picture of what it is doing in this area and why it is doing it.