taking with respect to the (b) (1), or national security, exemption, suggests that we are likely to experience a slow but steady leakage of information whose ultimate consequences are extremely hazardous. A part of this problem is one of a trend of judicial decisions, particularly in the District of Columbia Circuit and district court, that are making it increasingly difficult for the Government to maintain that exemption without disclosing considerable information in the course of litigation. In some cases—and this is a particularly acute problem because of the nature of our information-in some cases this involves almost as much information as the release of the disputed records might reveal.

The singular nature of NSA's activities makes even the process or responding to FOIA requests highly risky when those requests touch upon intelligence information, as most of the ones we receive do. It is a well known fact that NSA has only a single source of the intelligence information it produces, namely the interception of signals and communications. The fundamental security tenet of the signals intelligence process is that we must keep in strictest secrecy the identity of the Agency's targets, the degree of our ability to intercept those targets, the extent and nature of our ability to handle large masses of information and retrieve that which is of intelligence interest, and of course, the extent of any cryptanalytic success that we may have. The disclosure of any of these items of information we believe would encourage intelligence targets to take countermeasures, with the likely result that the United States would be denied valuable intelligence.

The nature of the product we have, and the unique, singular source from which it comes, makes particularly acute the problems to which Congressman Ashbrook alluded before; namely, that even if we apply ourselves to the process of sanitization and produce something which to the ordinary bystander conveys little or no intelligence information, in the hands of a knowledgeable person, for example, a person who was a party to an intercepted communication, the information that is left behind may render the process of sanitization unavailing. It is for this reason that we have had to take the position repeatedly in response to FOIA requests that there are no segregable portions of the documents in question and we cannot release any information whatsoever about them.

Even taking that position, the very process of responding to Freedom of Information Act requests seeking information contained in intelligence records poses problems for us, and in some cases the mere admission that NSA possesses information on a particular subject described in a request is enough to disclose classified information about intercept activities. Significant information about a withheld record may be disclosed merely by citing the statutory basis on which we are withholding it; 18 U.S.C. 798 reveals, if we cite it in a response, that the information being withheld was obtained by NSA from foreign communications, since that is the only subject matter protected from disclousure under that statute.

In other cases, admitting the existence of information in our files is tantamount to admitting or identifying the source, as, for example, when a person who requests NSA for records pertaining to himself is known to have engaged in only very limited foreign contacts. In cases of that sort, even admitting that we have information that refers to a named individual, put together with the fact that we are in the

business of intercepting foreign communications, may be a tipoff to a foreign intelligence service that we have in fact intercepted communications involving that individual, or to or from that individual to a specific foreign government.

Even the fact that we rely on section 403 (d) (3) of the National Security Act, the provision giving the Director of Central Intelligence responsibility to protect sources and methods, may convey information to a trained observer since, in our case, the intelligence sources and methods to which the statute apply could only be communications intelligence sources and methods.

Coupled with these problems of responding to a request even in the most conclusory and uncommunicative terms are the problems of the development of the law, in which the courts have exerted steadily increasing pressure to disclose in the public record more and more information about the documents being withheld. Information frequently is sought as to the dates of the withheld documents, the numbers of such documents, the number of pages in each of the documents, and each of these items, although possibly innocuous when applied to some other kind of intelligence information, can in particular cases be a tipoff as to the identity of the communications information possessed by the agency. Merely citing the executive order on which the original classification was based is a tipoff as to the general time frame in which the information was obtained, since there have been a succession of such executive orders over time.

The problems to which I have alluded are problems which arise in almost every kind of request, and they arise again and again in the first person requests where someone asks for whatever information we may have concerning himself. Other kinds of requests pose even more acute problems. For example, we get requests that pinpoint a category of intercepted communications or signals, that ask for communications sent on a particular day or between particular individuals or organizations, or that seek communications of a particular foreign government regarding a specific subject or incident. Such requests are by no means rare, and they appear to be in the process of becoming more frequent.

In cases like that, if we were to respond that we had records responsive to the request, even without releasing them, we would be as much as admitting the fact of the interception implied in the request. We are constrained by 18 U.S.C. 798 not to do this. On the other hand, it is out of the question to untruthfully deny the existence of records. We are required under the FOIA to make some kind of response, which leaves us with the only choice of refusing to confirm or deny the existence of information responsive to the request.

This response is necessary because if we admitted the existence of records in cases where we actually had them and they were responsive to the request, we would be revealing classified information. If we took the tack of refusing to answer in those cases but truthfully denying whenever we didn't have the records, whatever approach we took in the case where we did have records would be proved ineffective.

Even under our present system there is good reason to believe that many requesters assume whenever we refuse to confirm the existence or nonexistence of a record that such refusal is tantamount to an affirmation. We are concerned by the slow but steady cumulative buildup

of a mosaic of information about the Agency's communications intercept activities that a knowledgeable and sophisticated foreign observer could be deriving even from these shreds of what the requester presumably perceives as noninformation in the response from the agency. In addition, NSA takes seriously its obligation under the act to release as much requested information as possible, subject to the statutory exemptions. Consequently, we resort as infrequently as possible to a refusal to confirm or deny the existence of records. This means that someone must make a judgment in each particular case whether admitting that we have records in some way responsive to the request is likely to reveal sensitive information about sources and methods. These judgments are made on the basis of what we think generally would be known about the subject matter of the request, and there is always the possibility that the judgment is incorrect. If we get a request for information about a well-known, internationally active and traveled public figure, we may feel that we can confirm that we have some records that are either to or from that individual or mention that individual because of the multiple possibilities of intercept channels and foreign targets that could be involved. On the other hand, there is a very high risk that in making those judgments, we are less knowledgeable, less sophisticated than the requester or any foreign government that may be using this information and that we are making the kind of mistakes that Mr. Carlucci referred to in his testimony. There is another problem within NSA, also alluded to in Mr. Carlucci's testimony, inherent in the very act of processing these requests. As I indicated before, we consider ourselves obligated to make a reasonable search in good faith for the records, even though we may be virtually certain from the outset that whatever we find will be exempt from disclosure. These searches cut across our own system of compartmentation and across the principle of need to know, and result frequently in extremely sensitive information having a much wider distribution within the agency than it would under any other set of circumstances.

There is no evidence to date that this phenomon has resulted in any serious security breach, but the basic cornerstone of good security practice is to limit information to as few people as possible, and I think one has to take it for granted that spreading sensitive information around unnecessarily is a bad security practice and implies various security hazards.

All these problems that exist at the initial processing stage within the agency are magnified manyfold when an FOIA requester files suit in the district court, challenging the agency's refusal to disclose information. To date, no communications intelligence record withheld under either the (b) (3) or the (b) (1) exemption has been ordered released by a court. Subjecting such records to processing and litigation under FOI, consequently, is as futile as it is risky, and we certainly hope that it will continue to be futile, because if it doesn't, we are in very serious trouble with respect to the disclosure of sensitive sources and methods.

Under the Freedom of Information Act, the court has a de nova review function in considering the validity of FOIA exemptions and can examine classified records in camera to determine the propriety of withholding national security information. The legislative history

of the 1974 amendments which expand the power of the court in the area of de novo review and in camera inspection is not entirely clear, and there is great dispute in the law today as to precisely how far the District Court is intended to go and in what circumstances.

The courts, particularly in the District of Columbia, where most significant FOIA litigation takes place, are taking an increasingly expansive view of their role in de novo review of FOIA exemptions, and as I said before, are demanding more and more information about the records withheld. There is an increasing burden on the government to prove the validity of security classification and to convince the judge that if the judge were the original classifying official, he would have assigned the same classification. In many cases such proof is demanded on the public record, which exerts on the intelligence agency heavy pressure to disclose increasing amounts of information in an attempt to accomplish a task which is inherently impossible, that is to describe publicly the classified information and the damage that would result from its release, because a full public description of the information in most cases is tantamount to releasing the information itself.

Now, in my statement on pages 18 and 19, I describe some of the recent cases in the District of Columbia, including the recent decision in Ray v. Turner and the case of Shapp v. Turner, to which Mr. Carlucci alluded, and in which the U.S. Court of Appeals for the District of Columbia circuit, appeared to take a somewhat more expansive view of the role of the district court in reviewing both (b)(1) exemptions and (b) (3) exemptions in which the withholding statute is one intended to protect national security, such as Public Law 86-36. The concurring opinion in that case, the Ray case, by Chief Judge Wright, would carry the court's role a step further. Judge Wright would have the court, in addition to determining de novo the application of the exemption, require in camera submissions almost invariably as a check against agency abuse of the in camera affidavit process, would require the agency to explain why the information contained in its in camera submission should not have been included in a public affidavit, and would make available to all parties any portions of the in camera affidavits that the court determines, after full consideration of the Agency's arguments, do not warrant a protective order.

This procedure, which was not adopted by the majority of the court in that case, was nonetheless identical to that followed by one of the judges of the district court in a recent decision in Joan Baez v. National Security Agency, a case in which we filed an in camera affidavit explaining in considerable detail why the records that were being withheld, if disclosed, would cause damage to the national security. The district judge described himself as unconvinced by the Agency's affidavit, and ordered the affidavit to be released on the public record. This has not yet occurred because the Government has filed a motion for reconsideration which is still pending before the court. If it does, the damage to the national security from releasing that affidavit, in my opinion, will be significantly greater than would have occurred if the records themselves had been released in the first place, because the records require analysis and construction. The affidavit does that job very clearly and explicitly for the foreign intelligence services that might be interested. It tells them precisely what conclusions they

56-940 O 80 - 4

should draw from those records about our signals intelligence capabilities.

This kind of decision, obviously, makes us very nervous about the future of FOIA litigation involving our materials. These judicial trends that I have described, if continued, portend serious difficulties for NSA in maintaining secrecy about its communications intelligence activities. In virtually every case that we litigate, the rationale for withholding records is precisely the same: the record is either an intercepted communication or an intelligence report that describes an intercepted communication in terms that leave no doubt as to who the parties were, what the date was, and the route over which it was sent. The judgment that this information would be damaging to the national security if disclosed is an informed professional judgment based on extensive experience with the signals intelligence process. Frequently what it reflects is an understanding of what the United States could do with similar information about foreign signals intelligence activities in reconstructing the capabilities of foreign signals intelligenc services.

As indicated by experience in the Joan Baez case, however, it is difficult to communicate to judges who are not intimately familiar with the signals intelligence process the basis on which these judgments are made. The courts are confronted with able attorneys for FOIA requesters, who carefully marshall every scrap of information about NSA's activities, and then argue, on the basis of what they assert is already publicly known, that the additional disclosure of the requested information cannot be harmful. Frequently what they present to the court are surmises from the press, scraps of information that are not entirely accurate, but it is difficult in many cases for the courts to discern the difference between official release and press hypothesis, and this is an attractive argument, apparently, to the courts.

Mr. MURPHY. Mr. Silver, I am going to interrupt you at this point. You are almost through, I notice. You have just got another page or two.

The proposed amendment that Mr. Burlison requested, have you read it and studied it?

Mr. SILVER. I am not sure that I have seen the most recent version. Mr. MURPHY. I would like for you to take a copy of it to see what your legal department may want to add to it or delete from it, and let me just ask one question, and then I yield to Mr. McClory because we have a vote on the floor.

Has NSA received FOIA requests from foreign governments or aliens who seek to discover the information you have been talking about, your signals intelligence.

Mr. SILVER. I would have to give you the same answer that Mr. Carlucci did, that we have no way of knowing who was behind the FOIA requests. To my knowledge we have not received any that openly come from foreign governments or foreign intelligence services, but it is entirely possible that foreign governments are interested in responses to requests that we have received.

Mr. MCCLORY. You are only engaged in getting national security information, aren't you?

Mr. SILVER. Yes, sir.