PROVIDENCE JOURNAL COMPANY and Gerald M. Carbone, Plaintiffs, Appellees, v. UNITED STATES DEPARTMENT OF the ARMY, Defendant, Appellant, 981 F.2d 552

CYR, Circuit Judge. This appeal is taken from a district court judgment directing the United States Department of the Army (“Army”) to disclose to the Providence Journal Company (“Journal”), pursuant to a Freedom of Information Act (“FOIA”) request, numerous documents relating to an internal criminal investigation into allegations against six officers of the Rhode Island National Guard (“RING”). The Army contends that the documents are protected from compelled disclosure under three FOIA exemptions.

I BACKGROUND During 1988, the Office of the Inspector General of the Army (“IG”) received four anonymous letters implicating six RING officers in alleged misconduct punishable either by internal disciplinary action or by court-martial under the Uniform Code of Military Justice. See 10 U.S.C. §§ 801-946 (1985 & Supp.1992). The Army Vice Chief of Staff (“VCOS”) directed the IG to investigate the charges against two “senior” officers and to submit a report to the Army officer (“Army command”) invested with the authority to determine whether either disciplinary action or court-martial was warranted.

The allegations against the four junior officers were referred to the National Guard Bureau. In order to foster cooperation and curb possible fears of reprisal or harassment, the IG’s office, which has no subpoena power, promises confidentiality — as to both witness identity and statement content— “to the maximum extent possible, particularly when it is specifically requested.” Department of Army Regulation (“DAR”) 20-1, ¶ l-15a. The IG interviewed twenty-seven witnesses in the course of the investigation. Three witnesses waived their right to confidentiality.

In December 1989, the IG submitted a report (“IG Report”), which was “approved” by the Army VCOS. Army regulations provide that “approval” of an IG report does not connote official Army adoption of its findings or recommendations. DAR 20-1, 11 3-lc. The record reveals no further Army action on the IG Report. In due course, the Journal and one of its reporters filed an FOIA request for “all documents pertaining to the Inspector General’s investigation of the Rhode Island National Guard.” See 5 U.S.C. § 552 (1990).

The Army released a redacted version of the IG Report, withholding several exhibits in reliance on four FOIA exemptions. See id. §§ 552(b)(5) (exemption for predecisional intra-agency memoranda), (6), (7)(C) (exemptions to safeguard against unwarranted invasions of privacy), and (7)(D) (exemption for information provided by a “confidential source”). Following an unsuccessful administrative appeal to the Army General Counsel, the Journal filed suit in the United States District Court for the District of Rhode Island to compel disclosure of the unredacted documents pursuant to 5 U.S.C. § 552(a)(4)(B).

The parties filed cross-motions for

summary judgment. The district court directed the Army to submit a so-called Vaughn Index, see Vaughn v. Rosen, 484 F.2d 820, 824 (D.C.Cir.1973), cert. denied, 415 U.S. 977, 94 S.Ct. 1564, 39 L.Ed.2d 873 (1974), which lists the precise grounds for the Army’s exemption claims with respect to each redaction or withheld document, as follows: A.

IG Report 1 Identity of the six RING officers who were targets of the investigation [Exemptions 6 & 7(C)]; 2,6 IG’s conclusions as to whether each allegation was substantiated or unsubstantiated [Exemption 5]; 3,5 IG’s synopsis of each allegation and findings of fact [Exemptions 5 & 7(D)]; 4 Statements provided by confidential and non-confidential witnesses [Exemptions 5 & 7(D)]; 7 IG’s final recommendations regarding further disciplinary action [Exemption 5]; B. Full transcript of statement by Nonconfidential source [Exemptions 5 & 7(D)]; C-E.

Internal memoranda and directives between Army VCOS and IG’s Office [Exemptions 6, 7(C) & 7(D)]; F-I. Four anonymous letters [Exemptions 6, 7(C) & 7(D)]; J. Travel vouchers [Exemptions 6, 7(C) & 7(

D) l Following an in camera inspection of the unredacted documents, the district court granted partial

summary judgment and directed the Army to release the entire IG Report, excepting only the names (and other identifying information) of the confidential sources (Vaughn Index §§ A3, A4, AS) and the various intra-agency memoranda (Vaughn Index §§ C-E). II DISCUSSION The FOIA was designed to expose the operations of federal agencies to public scrutiny without endangering efficient administration, as a means of deterring the development and application of a body of “secret law.” See Department of Air Force v. Rose, 425 U.S. 352, 360-61, 96 S.Ct. 1592, 599, 48 L.Ed.2d 11 (1976); NLRB v.

Sears, Roebuck & Co., 421 U.S. 132, 153, 95 S.Ct. 1504, 1518, 44 L.Ed.2d 29 (1975). As the FOIA presumes public entitlement to agency information, an agency which would withhold information must establish its right to an FOIA exemption. See 5 U.S.C. § 552(a)(4)(B). The district court must make a de novo determination as to the validity of the agency’s exemption claim. See Department of Justice v. Reporters Comm. for Freedom of the Press, 489 U.S. 749, 755, 109 S.Ct. 1468, 1472, 103 L.Ed.2d 774 (1989). FOIA exemptions are construed narrowly, Department of Justice v.

Julian, 486 U.S. 1, 8, 108 S.Ct. 1606, 1611, 100 L.Ed.2d 1 (1988); Curran v. Department of Justice, 813 F.2d 473, 473-74 (1st Cir.1987), and any “[djoubts are customarily to be resolved in favor of openness.” Irons v. FBI, 811 F.2d 681, 685 (1st Cir.1987) [hereinafter “Irons I”]. A.

Exemption 5 With respect to the IG Report’s “subjective” evaluation of the evidence against the two senior RING officers, as well as the IG’s recommendations to the Army VCOS, the Army asserts a claim under Exemption 5 which prohibits compelled disclosure of “inter-agency or intra-agency memorandums or letters which would not be available by law to a party other than an agency in litigation with the agency.” 5 U.S.C. § 552(b)(5).

Agency documents which would not be obtainable by a private litigant in an action against the agency under normal discovery rules (e.g., attorney-client, work-product, executive privilege) are protected from disclosure under Exemption 5. United States v. Weber Aircraft Corp., 465 U.S. 792, 799, 104 S.Ct. 1488, 1492, 79 L.Ed.2d 814 (1984); EPA v. Mink, 410 U.S. 73, 86, 93 S.Ct. 827, 835, 35 L.Ed.2d 119 (1973).

The Army relies on the executive or “deliberative process” privilege, see, e.g., id. at 85-86, 93 S.Ct. at 835 (1973) (national security memo on nuclear testing prepared for President), which is designed to safeguard and promote agency decisionmaking processes in at least three ways: [I]t serves to assure that subordinates within an agency will feel free to provide the decisionmaker with their uninhibited opinions and recommendations without fear of later being subject to public ridicule or criticism; to protect against premature disclosure of proposed policies before they have been finally formulated or adopted; and to protect against confusing the issues and misleading the public by dissemination of documents suggesting reasons and rationales for a course of action which were not in fact the ultimate reasons for the agency’s action.

Coastal States Gas Corp. v. Department of Energy, 617 F.2d 854, 866 (D.C.Cir.1980); see also Schell v. Department of Health and Human Servs., 843 F.2d 933, 939 (6th Cir.1988). After considering any potential impact public disclosure might have on the employee-advisor, the agency decisionmaker, and the public, the court should construe Exemption 5 as narrowly as is “consistent with efficient Government operation.” Mink, 410 U.S. at 89, 93 S.Ct. at 837 (citation omitted). Normally, a document will qualify for protection under Exemption 5 if it is both “predecisional” and “deliberative.” See Dow Jones & Co. v.

Department of Justice, 908 F.2d 1006, 1008-09 (D.C.Cir.1990). 1. “Predecisional Document” Test A document will be considered “predecisional” if the agency can- (

i) pinpoint the specific agency decision to which the document correlates, Paisley v. CIA, 712 F.2d 686, 698 (D.C.Cir.1983), (ii) establish that its author prepared the document for the purpose of assisting the agency official charged with making the agency decision, Renegotiation Bd. v. Grumman Aircraft Eng’g Corp., 421 U.S. 168, 184, 95 S.Ct. 1491, 1500, 44 L.Ed.2d 57 (1975); Hopkins v.

Department of Hous. and Urban Dev., 929 F.2d 81, 84 (2d Cir.1991); Coastal States, 617 F.2d at 866, and (iii) verify that the document “precedes, in temporal sequence, the ‘decision’ to which it relates.” Senate of Puerto Rico v. Department of Justice, 823 F.2d 574, 585 (D.C.Cir.1987). The Journal concedes that the Army VCOS ordered the IG to conduct the preliminary criminal investigation and that Army command, not the IG, is the final decisionmaker as to whether there is to be any further disciplinary or prosecuto-rial action against the RING officers.

See Rules for Court-Martial 306(a) (“Each commander has discretion to dispose of offenses _”); Hopkins, 929 F.2d at 85 (document “predecisional” if its author “lacked any authority to take final agency action”). Thus, the IG Report would be a predeci-sional document. The Journal argues nonetheless that Army command implicitly adopted the IG Report by its apparent failure to take any action within a reasonable time after issuance, thereby disentitling the IG’s recommendations to “predecisional” status under Exemption 5.

The Journal contends that its “implied adoption” theory is necessary to prevent an agency’s use of its own inaction as an absolute shield from compelled FOIA disclosure of the results of any internal investigation. The “implied adoption” theory is neither supported by the plain language of Exemption 5 nor the related caselaw, and would disserve the recognized aims of Exemption 5. Express adoption of a predecisional document is a prerequisite to an agency waiver under Exemption 5.

See, e.g., Sears, 421 U.S. at 161, 95 S.Ct. at 1521 (agency must “expressly ... adopt or incorporate [predecisional document] by reference” in final decision); Ahearn v. United States Army Materials & Mechanics Research Ctr., 580 F.Supp. 1405, 1407 (D.Mass.1984) (same). Courts consistently have refused to infer agency adoption based on mere agency inaction. See, e.g., Brinton v. Department of State, 686 F.2d 600, 605 (D.C.Cir.1980) (age or length of retention of predecisional document irrelevant to question of agency “adoption”), cert. denied, 452 U.S. 905, 101 S.Ct. 3030, 69 L.Ed.2d 405 (1981); Ashley v.

Department of Labor, 589 F.Supp. 901, 908 (D.D.C.1983) (no implied adoption “even if a disputed document is several years old ... [and] has not yet produced a[n] [anticipated] final decision”). The proposed “implied adoption” rule would undermine Exemption 5 by inhibiting the free exchange of views within an agency. Agency advisors responding to supervisory directives might be less forthcoming with their advice lest their recommendations be exposed to public scrutiny in the event final agency action is not promptly taken. See Access Reports v.

Department of Justice, 926 F.2d 1192, 1196 (D.C.Cir.1991) (“At the time of writing the author could not know whether the decisionmak-ing process would lead to a clear decision, establishing the privilege, or fizzle, defeating it. Hedging his bets, he would be drawn into precisely the caution ... that the exemption seeks to render unnecessary.”); Schell, 843 F.2d at 941 (same). Especially is this true where, as here, one viable agency option is to take no final “action” on the IG’s recommendations. See Rules for Court-Martial 306(c)(1) (“A commander may decide to take no action on an offense.

If charges have been preferred, they may be dismissed.”). Accordingly, we decline to depart from the established view that an agency may meet its burden of proof under the “predecisional document” test by demonstrating that the preparer was not the final decisionmaker and that the contents confirm that the document was originated to facilitate an identifiable final agency decision. See Mobil Oil Corp. v.

EPA, 879 F.2d 698, 703 (9th Cir.1989) (agency asserting Exemption 5 claim need not demonstrate differences between contents of deliberative document and final agency decision). 2. “Deliberative Document” Test A “predecisional” document may still not “fall within the confines of Exemption 5 if it is not part of the ‘deliberative process.’ ” Formaldehyde Inst. v. Department of Health and Human Servs., 889 F.2d 1118, 1121 (D.C.1989); see also Access Reports, 926 F.2d at 1195 (document must “reflect[ ] the give-and-take of the consultative process”) (citation omitted).

The Army asserts that two features of the IG Report contributed to the deliberative process:

(1) Vaughn Index §§ A2, A6 and A7, conveying the IG’s views as to whether the allegations were substantiated, as well as the IG’s recommendations relating to any appropriate agency action, and

(2) Vaughn Index §§ A3 and A5, which include the IG’s findings of fact and summaries in support of the IG’s recommendations. The district court ruled that: the [IG’s] investigatory report was not a deliberative policy-making document. The investigation concerned factual allegations against high-ranking officials. This is not agency policy in the same vein as Mink, supra, where reports were prepared for the President on the advisability of underground nuclear testing.

If the [IG] report concerned broader issues—if it was a report of general recommendations on disciplining superior officers—the situation would be different. However, this report is factually specific; it does not reflect “agency give- and-take—of the deliberative process— by which the decision itself is made.” Providence Journal v. Department of Army, 781 F.Supp. 878, 885 (D.R.I.1991) (citing Weber Aircraft, 465 U.S. 792, 104 S.Ct. 1488, 79 L.Ed.2d 814 (1984); Cooper v. Department of Navy, 558 F.2d 274 (5th Cir.1977)) (emphasis added) (other citations omitted).

A predecisional document will qualify as “deliberative” provided it (

i) formed an essential link in a specified consultative process, (ii) “reflect[s] the personal opinions of the writer rather than the policy of the agency,” and (iii) if released, would “inaccurately reflect or prematurely disclose the views of the agency.” National Wildlife Fed’n v. Forest Serv., 861 F.2d 1114, 1118-19 (9th Cir.1988); see also SafeCard Servs., Inc. v. SEC, 926 F.2d 1197, 1204 (D.C.Cir.1991) (agency must show the decisional “context” of the document within the process used to reach determinations “like those in issue”); cf.

Senate of Puerto Rico, 823 F.2d at 585-86 (agency bears burden of establishing “what deliberative process is involved, and the role played by the documents in the course of that process”) (citation omitted). Even where expressions of personal opinion generally render a document “deliberative,” however, segregable factual portions of the document might still be subject to compelled disclosure if, for example, they are not so “inextricably intertwined” with the deliberative material that their disclosure would compromise the private remainder of the documents.

See Mink, 410 U.S. at 92, 93 S.Ct. at 838. a. “Consultative Process” We find no authority for the suggested distinction between “reports of general recommendations on disciplining superior officers” and “factual” reports prepared in the course of internal disciplinary investigations against particular individuals. See, e.g., Renegotiation Bd., 421 U.S. at 184, 95 S.Ct. at 1500 (agency deliberations preceding adjudicative decision involving specified persons implicate Exemption 5); National Wildlife, 861 F.2d at 1118 (Exemption 5 not limited to consultations over official “policy”); Brockway v.

Department of Air Force, 518 F.2d 1184, 1192 (8th Cir.1975) (Exemption 5 extends beyond “policy” memoranda to include all documents not discoverable in litigation with agency)', see also, e.g., Swisher v. Department of Air Force, 495 F.Supp. 337 (W.D.Mo.1980), aff'd, 660 F.2d 369 (8th Cir.1981) (IG Report constitutes “deliberative” document); American Fed’n of Gov’t Employees v. Department of Army, 441 F.Supp. 1308 (D.D.C.1977) (same). Rather, the appropriate judicial inquiry is whether the agency document was prepared to facilitate and inform a final decision or deliberative function entrusted to the agency.

See, e.g., Russell v. Department of Air Force, 682 F.2d 1045, 1046-48 (D.C.Cir.1982) (editorial review process used by Office of Air Force History to prepare historical document on use of Agent Orange during Vietnam war constitutes deliberative agency function). As Army command controls the agency decision whether Army personnel are to be disciplined for alleged misconduct, or prosecuted under the Uniform Code of Military Justice for alleged criminal activity, its deliberative task is no less an agency function than the formulation or promulgation of agency disciplinary policy.

As with other discretionary prosecutorial decisions, many considerations contribute to the final determination by Army command, including the rank of the investigated officers, the seriousness of the allegations, the overall reliability of the evidence, the relative appropriateness of the available forms of remediation, and any special mitigating circumstances. Cf. Senate of Puerto Rico, 823 F.2d at 585 n. 38 (“[T]he process leading to a decision to initiate, or forego, prosecution is squarely within the scope of the privi-lege_”).

It is not surprising, therefore, that the Army has in place a confidential consultative process to ensure maximum input from the chain of command concerning the need for further action.

See Russell, 682 F.2d at 1048 (agency has “much at stake” in “candid consideration” where it must be prepared to “stand by its [final decision] in the public forum, and in light of the possibility of ... litigation ... perhaps in the judicial forum as well”). b. “Essential” to Consultative Process Neither can we agree that the primary function of the IG Report was to convey raw evidence or data discovered during the investigation and that the IG’s recommendations were peripheral or gratuitous.

Schell, 843 F.2d at 940 (court must determine whether document was “essential” or merely a “peripheral item which just ‘beefs up’ a position with cumulative materials”) (citation omitted). We think it is clear that the recommendations made by the IG — the agency official with the investigative expertise and the greatest familiarity with the first-hand evidence — are highly important to Army command even though it is not obligated in the final analysis to credit the IG’s recommendations.

See, e.g., Hopkins, 929 F.2d at 85 (HUD inspector reports contain “recommendations to higher officials that various agency actions should be taken.”); Formaldehyde, 889 F.2d at 1125 (reliance on temporary consultants’ opinion often necessary); Schell, 843 F.2d at 942 (“It is the free flow of advice, rather than the value of any particular piece of information, that Exemption 5 seeks to protect.”). We cannot say that the IG’s recommendations were in any sense either merely cumulative or peripheral.

We conclude, at a minimum, therefore, that a significant portion of the IG Report (Vaughn Index §§ A2, A6, and A7) was “essential” to the consultative process within the agency. c. Premature Disclosure of IG’s “Personal Opinions” Nor is the chilling effect on candid advice from agency subordinates, which Exemption 5 was designed to mitigate, significantly diminished merely by reason of the fact that the subordinates’ recommendations relate to the appropriateness of disciplinary action against particular individu-ais.

A subordinate agency advisor may have more cause for concern about public disclosure of disciplinary recommendations involving high-level agency officials, since there may be a real or perceived risk of retaliation from a vindictive official who is the target of the advisor’s findings or recommendations. Cf. Cooper, 558 F.2d at 277 (“[S]ervice people are human, too: they fear disciplinary action, work and hope for promotion, possess loyalties and ties of friendship to people and organizations, [and] dislike speculating to the derogation of others’ reputations_”).

Army command is not required to accept the IG’s recommendations. Indeed, command already may have exercised its prerogative to take no further action on these allegations, for reasons entirely unrelated to the grounds espoused in the IG Report.

Accordingly, since public release of the recommendatory sections in the IG Report would either “inaccurately reflect or prematurely disclose the views of the agency,” the Army may not be required to reveal any information referenced in Vaughn Index §§ A2, A6, or A7. d. “Inextricably Intertwined” Fact-Oriented Material The Army contends that Vaughn Index §§ A3 and A5, conveying the IG’s conclusions as to the facts revealed by the evidence discovered during the investigation, should be exempt because the conclusions are so “inextricably intertwined” with the IG’s mental processes that their disclosure necessarily would reveal the substance of the IG’s recommendations.

See, e.g., Quarles v. Department of Navy, 893 F.2d 390, 392-93 (D.C.Cir.1990) (cost estimates derive from “complex set of judgments” by preparers); Russell, 682 F.2d at 1048 (historical facts essentially “interpretive” choices by reviewer); Swisher, 495 F.Supp. at 340 (IG’s investigative conclusions exempt); American Fed’n, 441 F.Supp. at 1313 (IG’s preliminary conclusions “play an integral

part in the consultative process”). The district court held that the IG’s evidentiary conclusions and rationale are not exempt from disclosure since “[t]he mere act of selecting facts for inclusion in a report does not make that report deliberative,” and “[sjimple judgment exercised in preparing the Report of Investigation does not equal deliberation.” Provi dence Journal, 781 F.Supp. at 885.

The Exemption 5 analysis employs a rough-hewn dichotomy between opinion and fact: whereas the purely recommenda-tory provisions in a deliberative predeci-sional document are exempt from compelled disclosure, “memoranda consisting only of compiled factual material or purely factual material contained in deliberative memoranda and severable from its context would generally be available for discovery....” Mink, 410 U.S. at 87-88, 93 S.Ct. at 836 (emphasis added); see also Hopkins, 929 F.2d at 85; Russell, 682 F.2d at 1048; Mead Data Cent., Inc. v. Department of Air Force, 566 F.2d 242, 256 (D.C.Cir.1977).

As the dichotomy between opinion and fact is not clear-cut, courts generally follow a “functional” approach in an attempt to determine “whether production of the contested document [or section] would be ‘injurious to the consultative functions of gov-ernment_ Mink, 410 U.S. at 87, 93 S.Ct. at 836 (citing Kaiser Aluminum & Chem. Corp. v. United States, 157 F.Supp. 939, 946, 141 Ct.Cl. 38 (1958)).

Even when requested material is found to be factual, the courts have held it exempt where they were convinced that disclosure “would expose an agency’s de-cisionmaking process in such a way as to discourage candid discussion within the agency and thereby undermine the agency’s ability to perform its functions.” Quarles, 893 F.2d at 392 (citation omitted); see also Access Reports, 926 F.2d at 1195 (central inquiry is whether disclosure would “discourage candid discussion within the agency.”) (citation omitted); Formaldehyde, 889 F.2d at 1123 (opinion-fact distinction is subordinate to inquiry concerning “effect of the materials’ release” on deliberative process).

Factual material should be considered segregable if it is not so “inextricably intertwined” with the deliberative material that its disclosure would “compromise the confidentiality of deliberative information that is entitled to protection.” Mink, 410 U.S. at 92, 93 S.Ct. at 838; see Hopkins, 929 F.2d at 85. While mere selection of the evidence deemed material to an agency decision may not implicate Exemption 5, disclosure of the IG’s findings of fact necessarily would reveal the opinion of the IG on the credibility and probity of the evidence relating to each allegation.

Findings of fact arrived at in the personnel management context reflect a significant degree of subjectivity. Our review of these Vaughn-indexed documents discloses instances of conflicting and inconsistent witness statements. The findings of fact in the IG Report necessarily were premised on an assessment and resolution of the relative credibility of these statements, as well as subjective judgments as to the probity of other evidence developed during the investigation. Cf. Playboy Enters., Inc. v.

Department of Justice, 677 F.2d 931, 935 (D.C.Cir.1982) (plaintiff did not “wish to probe the process whereby the task force assigned reliability or weight to specific evidence”). Revelation of the IG’s findings of fact undoubtedly would divulge the substance of the related recommendatory sections with which they comport. Accordingly, as we conclude that the recommenda-tory provisions in the IG Report are exempt from disclosure, the Army cannot be compelled to disclose the IG’s findings of fact in Vaughn Index §§ A3 and A5. B. Exemption 7(

D) The Army claims that the statements provided by twenty-eight (twenty-four solicited witnesses and four anonymous informants) of its thirty-one sources (Vaughn Index §§ A4, F-

I) are nonetheless protected under Exemption 7(D), which shields from compelled disclosure records and information compiled for law enforcement purposes, but only to the extent that the[ir] production ... could reasonably be expected to disclose the identity of a confidential source and, in the case of a record or information compiled by criminal law enforcement authority in the course of a criminal investigation ... information furnished by a confidential source. (Emphasis added.) Exemption 7 was intended to avert the “drying-up” of sources of information necessary to conduct criminal investigations. Irons v.

FBI, 880 F.2d 1446, 1450-51 (1st Cir.1989) [hereinafter “Irons II”]. An agency claiming the right to decline disclosure on the basis of Exemption 7(

D) must demonstrate that the particular document was compiled for “law enforcement purposes” from information provided by a “confidential source.” See Curran, 813 F.2d at 473-74 (unlike certain other FOIA exemptions, once both elements of a 7(

D) exemption are established, the court should not engage in a balancing of interests); Brant Constr. Co. v. EPA, 778 F.2d 1258, 1262-63 (7th Cir.1985) (same). The Journal challenges only the “confidentiality” of the Army’s sources. Document confidentiality depends not on the contents but on the terms and circumstances under which the information was acquired by the agency. See Irons I, 811 F.2d at 685; Johnson v. Department of Justice, 739 F.2d 1514, 1517 (10th Cir.1984); see also Irons II, 880 F.2d at 1448 (“confidential” does not mean “secret” information, but information “provided in confidence”).

A confidential source is one who “ ‘provide[s] information under an express assurance of confidentiality or in circumstances from which such an assurance could be reasonably inferred.’ ” Id. at 1447 (quoting S.Con.Rep. No. 1200, 93d Cong., 2d Sess. 13 (1974), U.S.Code Cong. & Admin.News 1974, 6267, 6285, 6291) (citation omitted). We discuss the two types of source statements requested by the Journal: (1) solicited statements from confidential sources (Vaughn Index § A4), and (2) unsolicited statements from confidential sources (Vaughn Index §§ F-I). 1.

Statements Solicited from Confidential Sources (Vaughn Index § A4) The district court found that twenty-four of the twenty-seven individuals from whom information was solicited by the IG qualified as “confidential” sources, since they accepted express agency assurances of confidentiality; hence their identities were protected from disclosure under the first clause of Exemption 7(D).

The district court nevertheless held that the contents of each statement solicited from these confidential sources must be disclosed because the information was not “furnished only by the confidential source.” See Providence Journal, 781 F.Supp. at 886-87. The court did not identify the nonconfidential sources to which it made reference, but presumably meant to include the anonymous authors of the four unsolicited letters which triggered the internal investigation, as well as the three sources who expressly declined assurances of confidentiality.

Although we agree with the district court that the identities of the twenty-four confidential sources are exempt from disclosure under the first clause of Exemption 7(D), we cannot agree that the contents of their statements are not shielded from disclosure under the second clause. The district court ruling relied entirely on the word “only”—appearing in the second clause of Exemption 7(D)—which Congress deleted in 1986 for the explicit purpose of clarifying the broad policy goals served by the second clause.

Irons I, 811 F.2d at 687 (“ ‘There should be no misunderstanding that ... [the 1986 modifications] are intended to broaden the reach of this exemption and to ease considerably a Federal law enforcement agency’s burden in invoking it.’ ”) (citing 199 Cong.Rec. S16504). Under amended Exemption 7(D), an agency may not be ordered to disclose information from a confidential source even if nonconfi-dential sources have provided the agency with the identical information.

The Journal nevertheless urges af-firmance of the district court ruling, on the ground that the Army should be required to prove that each individual witness either (1) initiated the request for confidentiality, or (2) articulated a legitimate reason for invoking confidentiality respecting statement content after the IG made the initial tender of confidentiality. Absent some such prophylactic rule, the Journal argues, a federal agency could insulate itself from legitimate FOIA disclosure requests merely by offering confidentiality to all sources, whether or not required or requested by the source.

The Journal cites no authority for its proposed rule, and sound policy considerations counsel against it. Muzzling law enforcement agencies in order to deter tenders of confidentiality likely would risk “drying-up” the flow of information from many wary witnesses with valuable information, especially sources who might reasonably expect that an agency would extend an offer of confidentiality if it were an available option.

Moreover, in circumstances where law enforcement officials solicit information pertaining to a criminal investigation, see supra note 13, absent evidence to the contrary the courts have inferred agency assurances of confidentiality notwithstanding agency silence. See, e.g., Dow Jones, 908 F.2d at 1010. A requirement that agency assurances of confidentiality be subjected to post hoc judicial evaluation as proposed by the Journal promises more mischief than benefit.

The task of evaluating the “legitimacy” of confidentiality claims and assurances on a witness-by-witness basis would not only be onerous but often fruitless. It would be rare that a source would be unable to advance some colorable basis for a confidentiality claim, given the subjective nature of witness concerns about possible retaliation.

Perhaps more importantly, mere awareness by potential sources that the agency’s assurances of “content” confidentiality would be subject to second-guessing by the courts (advice with which fairness would seem to require that an agency provide its potential sources in advance) frequently would mean that only the unwary would be inclined to provide information in an internal criminal investigation. We think it more fair and efficient that law enforcement agencies be allowed to continue to extend assurances of confidentiality to their sources, with the advice that confidentiality may be disclaimed.

In this manner, unfettered agency control and manipulation of Exemption 7(

D) protections can be minimized without jeopardizing valuable agency sources. As the procedure utilized by the Army met this standard, its twenty-four solicited statements are exempt from FOIA disclosure in their entirety under Exemption 7(D). 2. Unsolicited Anonymous Sources (Vaughn Index §§ F-

I) The district court found that the four anonymous letters were not protected from compelled disclosure by Exemption 7(

D) as there was no evidence that the letters were sent under implied assurances of confidentiality. The court identified two reasons for its ruling: (1) the letters may have been written by non-military personnel unfamiliar with the "obscure Army regulations” assuring confidentiality, and (2) copies of the letters were made available to other “disciplinary” officials, including Army generals and the Governor. Providence Journal, 781 F.Supp. at 887.

Given the obvious import of the 1986 amendments to Exemption 7(D), see supra note 14, and the consequent easing of the law enforcement agency’s burden of proof, we think the Tenth Circuit has articulated a sound rationale for determining whether the unsolicited information from these anonymous sources was provided under an implied assumption of confidentiality. See Johnson, 739 F.2d at 1517-18 (summarizing three-way circuit split and citing cases). Most people would assume that the information they give to a criminal law enforcement official during a criminal investigation will be kept confidential.

However, situations may arise where it is unreasonable to make this assumption, and in the face of evidence to this effect in the record, a district court will not be precluded from so finding. Id. at 1518. Generally speaking, the circumstances in which these anonymous letters were submitted comport with a reasonable assumption of confidentiality on the part of the writers. Thus, [i]t is unrealistic to assume that a majority of persons reporting to an agency what they believe to be illegal or improper acts are legally sophisticated.

To the contrary, it is much more likely that they would not know the boundaries of the FOIA exemptions and, therefore, would not include in their initial communication to the agency an express request for confidentiality. They may be frightened, angry, or confused, and their immediate concerns do not include creating an evi-dentiary record to prove an assurance of confidentiality in anticipation of a potential FOIA request.

In cases involving unsolicited information from ostensibly confidential sources, the court should look to all factors ... [to determine] whether a request for confidentiality is implicit, i.e., that in light of the information and surrounding circumstances, the communication in all likelihood would not have been made if confidentiality had not been assured. Brant, 778 F.2d at 1263-64 (citation omitted) (emphasis in original) (distinguishing between confidentiality standards applicable to solicited and unsolicited sources).

At least in circumstances where the allegations might lead to court-martial proceedings, it is reasonable to infer, absent contrary evidence, that an anonymous source reasonably expected complete confidentiality. Evidence that might arguably support a contrary inference in the present case would be as follows: (1) the writers' decisions not to sign their names; (2) their expressed intention to provide copies to non-agency officials; and (3) any other intrinsic manifestations in the letters which might reflect their lack of concern about public dissemination of their letters.

Unlike Brant, which involved an identified unsolicited source, see id. at 1260-61, in the present ease the writers redacted their names. The Journal contends that the redaction of their names demonstrates the writers’ realization that the letters might be publicized, and their satisfaction that they had successfully excised all forms of identifying material. On the contrary, we believe that their determination to remain anonymous provides further reason for indulging the customary presumption that the information was provided under an implicit assurance of confidentiality. Exemption 7(

D) contains two independent safeguards against content disclosure. Under the first clause, there is to be no disclosure of information which would reveal the identity of the source. Under the much broader second clause, however, there is to be no disclosure of information offered in confidence, without regard to whether it would divulge the identity of the source. See Irons II, 880 F.2d at 1452 (“ ‘[Information furnished’ exemption [applies] irrespective of subsequent public identification of the source_”); Shaw v. FBI, 749 F.2d 58, 62 (D.C.Cir.1984) (unless second clause of Exemption 7(

D) protects information beyond that which would reveal the identity of the source, it is redundant); Johnson, 739 F.2d at 1517 (same). The congressional purpose underlying the blanket exemption in the second clause is readily apparent.

Although even a known source may not want the substance of the information made public, sources who choose to clothe themselves in anonymity most likely do so because they do “not want to have to rely upon the agency’s or the courts’ judgment that disclosure will not reveal [their] identity (which is of course the basis for a separate exemption — the first clause of Exemption 7(D) ...).” Shaw, 749 F.2d at 61.

Indeed, without knowing the identity of the source, and any other potential clues to the source’s identity which the details of the allegations might afford the target, often the court would be hard put to sift all identifying information from an anonymous letter. Over the long term, uncertainty about the sureness and consistency of this sort of post hoc judicial determination could affect the flow of important information from anonymous sources necessary to effective law enforcement.

Thus, if Exemption 7(D), clause 2, would preclude disclosure of statements solicited from confidential sources even though their names are redacted, it is not clear to us that the writer’s redaction of his or her name from an unsolicited letter, without more, gainsays the reasonableness of the normal presumption that the writer of the anonymous letter anticipated the maximum level of confidentiality which would be available to other confidential sources. Second, we can ascribe no controlling significance to the fact that the authors of three of the anonymous letters (Vaughn Index §§ F, G,

H) expressed their intention to provide copies to non-agency officials, such as the Governor of Rhode Island, who is vested with concurrent authority to pursue disciplinary action against RING personnel. See R.I.Gen.Laws § 30-2-1 (1982) (prescribing Governor’s statutory authority as commander-in-chief of RING).

Assuming the writers carried through with their stated intention to send duplicate letters, there is nothing in the record to suggest that the writers could not reasonably have expected comparable assurances of confidentiality from these non-agency officials, or that these officials dealt with the letters in a manner which might arguably render the reasonableness of the writers’ expectations suspect.

See Brant, 778 F.2d at 1264 (simultaneous submission of unsolicited letter to federal and state enforcement agencies did not undermine implied assurance of confidentiality, where “nothing in the record indicates ... that these [other] agencies did not treat the letter as confidential”). Finally, two letters (Vaughn Index §§ F,

I) contain explicit representations that the writers feared “reprisal” or “retribution” {e.g., loss of employment) in the event their statements were disclosed. See id. (court ought not dismiss as “a flight of fancy” an expressed fear of retaliation in unsolicited letter). Given the core function of Exemption 7(D), we believe that the flow of unsolicited information should not be jeopardized by risking exposure of the identities of sources through disclosure of the contents of their anonymous communications.

We therefore conclude that all twenty-eight confidential source statements are protected from compelled disclosure by Exemption 7(D). The three remaining source statements, however, are not “confidential,” and therefore are not protected from compelled disclosure under Exemption 7(D). C. Exemption 7(

C) The three “nonconfidential” source statements (Vaughn Index § A4) include explicit references to the names of the two senior RING officers.

The district court did not distinguish between substantiated allegations and unsubstantiated allegations, but ordered disclosure of the names of the two senior RING officers because (1) as “high-ranking” agency officials with substantial supervisory authority, the officers enjoyed a diminished privacy interest, (2) the allegations of criminal conduct implicated their official duties, which would “shed light” on RING performance, and (3) there is a countervailing public interest in monitoring RING performance, both as concerns the conduct of the target officers and the adequacy and comprehensiveness of the IG’s internal investigation.

Providence Journal, 781 F.Supp. at 882-84.

The Army counters that it may withhold these officers’ names under Exemption 7(C), which protects from compelled disclosure “records or information compiled for law enforcement purposes, but only to the extent that the[ir] production ... could reasonably be expected to constitute an unwarranted invasion of personal privacy.” (Emphasis added.) The Army did not appeal the district court ruling compelling disclosure of the three nonconfidential source statements relating to the substantiated allegations, perhaps because the Army understood that “all or much of this material may be independently protected by exemptions 5 or 7(D).” Brief for Appellant at 20 n. 16 (emphasis added).

As neither Exemption 5 nor Exemption 7(

D) is availing, however, the Army must disclose, in their entirety, the statements of the three noncon-fidential sources which relate to substantiated allegations. We confine the remainder of our discussion to the nonconfidential source statements relating to the “unsubstantiated” allegations against the two senior RING officers. Under our Exemption 7(

C) precedents, in order to determine whether disclosure might reasonably be expected to work an “unwarranted” invasion of privacy the court is required to balance the privacy interests of the targets of the criminal investigation against any public interest in the disclosure of their identities. See New England Apple Council v. Donovan, 725 F.2d 139, 143 (1st Cir.1984); Sands v. Murphy, 633 F.2d 968, 971 (1st Cir.1980).

A private individual who becomes the target of a law enforcement agency investigation, and whose alleged criminal conduct in no way reflects on the law enforcement agency’s performance, has a significant interest in preventing premature public disclosure of his or her identity under Exemption 7(C). See Reporters Comm., 489 U.S. at 765, 773, 109 S.Ct. at 1477, 1481 (information concerning private citizens “reveals little or nothing about an agency’s own conduct”); Nadler v.

Department of Justice, 955 F.2d 1479, 1490 (11th Cir.1992) (“Enabling the public to learn about the conduct of private citizens is not the type of public interest the FOIA was intended to serve.”); Hopkins, 929 F.2d at 88 (same). On the other hand, a federal government employee investigated for criminal misfeasance relating to the performance of official duties generally possesses a diminished privacy interest. See Stern v. FBI, 737 F.2d 84, 92 (D.C.Cir.1984) (high-level FBI official).

It is equally clear, however, that an internal criminal investigation would not invariably trigger FOIA disclosure of the identity of the targeted government employee: ‘One who serves his state or nation as a career public servant is not thereby stripped of every vestige of personal privacy, even with respect to the discharge of his official duties. Public identification ... could conceivably subject them to harassment and annoyance in the conduct of their official duties and in their private lives.’ New England Apple, 725 F.2d at 142 (citation omitted); see also Fund for Constitutional Gov’t v.

National Archives & Records Serv., 656 F.2d 856, 864 (D.C.Cir.1981). Therefore, we must determine appropriate guidelines for weighing the privacy interest remaining to these RING officers against the public interest in the disclosure of their identities. Public identification of the “targets of law enforcement investigations can subject those identified to embarrassment and potentially more serious reputational harm.” SafeCard, 926 F.2d at 1205 (citations omitted).

In virtually all cases, however, disclosure of the information adduced in an agency investigation serves the public interest at least to the extent that it sheds light on the agency’s performance of its official duties. Cf. Rose, 425 U.S. at 367-69, 96 S.Ct. at 1601-02 (noting public interest in administration of internal discipline as it reflects on military preparedness); see also New England Apple, 725 F.2d at 144 (“The public has a significant, enduring interest in remaining informed about actions taken by public officials in the course of their official duties.”).

The higher the rank of the public official alleged to have engaged in misconduct, the greater the legitimate public interest in disclosure is likely to be. Stern, 737 F.2d at 92, 94; see also Hale v. Department of Justice, 973 F.2d 894, 901 n. 8 (10th Cir.1992) (noting that governmental misconduct by “high ranking officers” may tip balance in favor of civil or public interest).

In the case of a low ranking agency official or employee, a rebuttable presumption may arise against compelled disclosure of allegations of misconduct which the agency investigation determines to have been “unsubstantiated,” but the case becomes more complicated if the target is an agency official of greater authority or importance. In particular, there may be a greater public interest in disclosure where the allegation — although determined unsubstantiated by the agency — may nevertheless be true, and may pose a serious threat to the public interest.

Or an “unsubstantiated” allegation may bear upon a claim, supported by independent evidence, that the investigating agency actively engaged in the concealment of the target official’s misconduct or otherwise failed to perform its mission. These considerations lend themselves to no mechanical rule of disclosure or non-disclosure. Nor, on the other side of the Exemption 7(

C) equation, can we prescribe a formula for measuring the impact of the privacy invasion resulting from disclosure. These and other relevant variables must be determined and weighed in light of the particular circumstances in each case. With these general considerations in mind, we turn to the particular facts before us. It is true, as the Army suggests, that the Journal neither alleged nor attempted to prove a cover-up in the IG’s investigation. At the same time, we think the invasion of privacy wrought by disclosure in this case is unusually slight.

The Army already has disclosed one of the two unsubstantiated allegations and the other is minimally invasive of privacy, containing as it does a rather blurred suggestion of possible impropriety. The unsubstantiated allegations are not of such an intimate nature that the disclosure of the target’s identity normally would be “unwarranted” even though the information might tangentially implicate the target’s performance of official duties, or the zeal or competence of the investigators. See, e.g., New England Apple, 725 F.2d at 143; see also Hunt v. FBI, 972 F.2d 286, 288-89 (9th Cir.1992).

Under all the circumstances, and eschewing the per se rules proposed by the parties, on balance we believe the Exemption 7(

C) analysis favors disclosure. Ill CONCLUSION The Army voluntarily disclosed redacted versions of the statements of its nonconfi-dential sources, redacting more than the officers’ names in some instances. See Vaughn Index A, p. 11, ¶ 8. The Army shall be required to release an unredacted version of the source statements appearing in the IG Report at p. 8, 1117; p. 11, ¶ 8; p. 14, H 10; p. 15, 11 3; and p. 17, II 5. The Army nonetheless may redact any reference to persons (other than the two senior RING officers) who acted as confidential sources and are identified as such in any nonconfidential source statement.

See, e.g., Vaughn Index A, p. 11, 11 8. The district court judgment is modified in accordance herewith, and affirmed as modified; no costs. .Throughout the opinion, relevant portions of the IG Report and exhibits are identified by reference to their Vaughn Index numbers. The Army did not appeal from the order to disclose, in their entirety, Vaughn Index B and J. See Providence Journal Co. v. Department of Army, 781 F.Supp. 878, 888-92 (D.R.I.1991) (Appendix A). . The Journal does not challenge the district court ruling relating to Vaughn Index C-E. .

Throughout the opinion, all citation references to agencies or departments are to United States agencies or departments, unless otherwise indicated. . The Journal suggests also that the Army’s earlier "approval” of the IG Report, coupled with the apparent inaction, signified official Army "adoption" of the IG Report.

Army Regulations provide, however, that "[w]hen an IG report is approved, conclusions and recommendations contained in the report do not constitute the directing authority’s decision nor an explanation of the decision unless specifically adopted as such in writing by the directing authority.” DAR 20-1, ¶ 3-lc (emphasis added); cf. Niemeier v. Watergate Special Prosecution Force, 565 F.2d 967, 973 (7th Cir.1977) (adoption requires something more than mere quotation, such as an affirmative manifestation that the predeci-sional document's conclusions are deemed “consistent" with final agency decision). .

The one decision cited in support of the theory, Washington Post Co. v. Department of Air Force, 617 F.Supp. 602, 605 (D.D.C.1985), is in-apposite, as it involved an express agency adoption of an IG report. . The Army argues that the Journal cannot assert its "implied adoption” claim because it failed to request disclosure of documents describing any final agency action following "approval” of the IG Report.

In our view, however, the initial Journal request, which sought “all documents pertaining to the Inspector General’s investigation of the Rhode Island National Guard," was broad enough to include any such written record of final action by the Army, if one exists. See McGehee v. CIA, 697 F.2d 1095, 1102-03 (D.C.Cir.1983) (once agency responds fully to FOIA requests, no continuing duty to disclose documents generated later; in general, prior to full compliance, there is an ongoing obligation to update disclosure).

For present purposes, we assume that no document evidencing final Army action has yet issued. .In contrast, express agency adoption represents a significant vindication of a subordinate advisor’s recommendation, posing little risk of retaliation or public embarrassment. See, e.g., SafeCard Servs., Inc. v. SEC, 926 F.2d 1197, 1204 (D.C.Cir.1991); see also Washington Post, 617 F.Supp. at 605. . None of the cases relied on by the Journal, or by the district court, supports a contrary result.

Weber Aircraft, 465 U.S. at 796, 104 S.Ct. at 1491 (Air Force waived Exemption 5 claim by voluntarily releasing entire record of collateral investigation of air crash); Playboy Enters., Inc. v.

Department of Justice, 677 F.2d 931, 935 (D.C.Cir.1982) (plaintiff sought disclosure of contents of witness statements only, but did not "'wish to probe the process whereby the task force assigned reliability or weight to specific evidence’ ”); Cooper, 558 F.2d at 279 (Navy investigative report of helicopter crash, which was primarily "fact-oriented,” with the expression of an opinion "incidental,” not entitled to blanket exemption; on remand, however, district court must scrutinize each

section of report to determine if its disclosure would "safeguard the consultative or decision-making process"); Brockway, 518 F.2d at 1185 (plaintiff sought only witness statements concerning air crash (not findings of fact)); Project on Military Procurement v.

Department of Navy, 710 F.Supp. 362, 367 (D.D.C.1989) (plaintiff entitled to waiver of fee on FOIA request; court does not reach merits of Navy's potential Exemption 5 claim, nor plaintiffs entitlement to disclosure); Washington Post Co., 617 F.Supp. at 605-07 (Air Force "waived” right to prevent disclosure of most fact-oriented portions of document either by express adoption of related recommendatory sections or by voluntary disclosure of summaries of more detailed fact-oriented sections; government failed to meet burden by providing "empirical support" that remaining fact-oriented sections were "inextricably intertwined” with exempted deliberative material).

Adams v. United States, 686 F.Supp. 417 (S.D.N.Y.1988), is the only case the Journal cites which is at all apposite. However, Adams stands on a mistaken premise. The Adams court held that the IG’s findings of fact were irrelevant to the deliberative process because the commanding officer was free to disregard the findings in making the final disciplinary decision, and that Exemption 5 applies "only insofar as [ ] disclosure might tend to expose the decisionmaker’s deliberative process.” Id. at 419-20 (citation omitted) (emphasis in original).

Exemption 5 protects the deliberative process, which necessarily involves at least two parties: the advisor and the decisionmaker. The fact that the decisionmaker may choose to disregard the IG's advisory findings does not alter the fact that (1) the IG, who had first-hand exposure to the witnesses and the evidence, is in the optimum position to make informed findings of fact; and (2) an informed final decision requires the IG’s candid assessment of witness demeanor and credibility. .

Even though the Army raised the issue before the district court, see Providence Journal, 781 F.Supp. at 889 (Appendix A), on appeal it apparently disclaims any contention that Vaughn Index § A4, objectively recounting the contents of the statements provided by the 27 solicited sources, is entitled to protection from FOIA disclosure under Exemption 5. See Playboy Enters., 677 F.2d at 935 (mere selection of facts for inclusion in report not “deliberative”). .

In some cases, a predecisional distillation of material facts from a larger public record may reveal the final decisionmaker’s mental processes by enabling public scrutiny of the information not relied on in arriving at the final agency decision. See, e.g., Russell, 682 F.2d at 1049 (comparison with final agency action would reveal what the agency thought were insignificant preliminary findings of fact); Montrose Chemical Corp. v. Train, 491 F.2d 63, 67-68 (D.C.Cir.1974) (same). But absent any documentation evidencing a final agency decision, we need not address this issue. .

For instance, disclosure of a finding that a high-ranking officer did or did not engage in particular conduct reveals the IG’s judgment as to whether the allegation was substantiated by the evidence. Were such findings subject to compelled disclosure in these circumstances, forthright findings of fact by agency subordinates, based on disputed evidence, would be harder to come by. See American Fed’n, 441 F.Supp. at 1313 (chilling effects of prospective disclosure are greatest in the case of an internal criminal investigation of agency personnel). .

The Journal seeks disclosure of the contents of the statements only, not the witnesses’ identities. . The Journal does not dispute that the requested documents satisfy the threshold criterion under Exemption 7(D), namely that the records were compiled by a criminal law enforcement authority in the course of a criminal investigation. See Curran, 813 F.2d at 475; Shaw v. FBI, 749 F.2d 58, 63 (D.C.Cir.1984) (agency must "identify ... a particular individual or a particular incident as the object of its investigation and ... the connection between that individual or incident and a ... violation of federal law."); cf.

Stern v. FBI, 737 F.2d 84, 89 (D.C.Cir.1984) (mere internal disciplinary proceeding not sufficient for Exemption 7(D); investigation must be targeted at specific person for actions punishable either by criminal or civil sanctions). For Exemption 7 purposes, Inspectors General are normally deemed "criminal law enforcement” agencies, Brant, 778 F.2d at 1265 (citing New England Apple Council v. Donovan, 725 F.2d 139 (1st Cir.1984)) (noting "the substantial similarities between the activities of the FBI and OIGs [Offices of Inspectors General]”).

These allegations exposed the RING officers to possible court-martial. . Prior to the 1986 amendment, § 552(b)(7)(

D) exempted investigatory records compiled for law enforcement purposes, but only to the extent that the production of such records would ... disclose the identity of a confidential source and, in the case of a record compiled by a criminal law enforcement authority in the course of a criminal investigation, or by an agency conducting a lawful national security intelligence investigation, confidential information furnished only by the confidential source. (Emphasis added.) The 1986 amendment eased the government’s burden of proof substantially.

For the phrase "would ... disclose,” it substituted the phrase “could reasonably be expected to disclose,” and it deleted the word "only" in the final clause. See supra p. 563; see also Irons I, 811 F.2d at 687. . The Journal acknowledges that the district court mistakenly relied on the unamended version of Exemption 7(D). It concedes as well that the district court’s finding that these twenty-four witnesses did receive express assurances of confidentiality is supportable. .

As the Army concedes, the statements provided by the three remaining sources, who expressly waived the IG's assurances of confidentiality, would in all likelihood not be protected from disclosure under Exemption 7(D). Exemption 7(

D) itself does not indicate what effect a witness's waiver of assurances of confidentiality might have on the agency’s power to shield the statement from FOIA disclosure. Nevertheless, since uncertainty about the precise scope of a waiver might "dry up” law enforcement sources, we consistently have refused to find an implied waiver where the subjective intent of the informant to relinquish confidentiality can be inferred only from ambiguous conduct, often occurring long after the informant provided the confidential information. See Irons I, 811 F.2d at 686; see also Parker v.

Department of Justice, 934 F.2d 375, 380-81 (D.C.Cir.1991). Although we need not resolve the question, recognition of an express waiver would not appear to pose any comparable risk of chilling “the flow of information to the law enforcement agency,” Irons II, 880 F.2d at 1449, since a source with sole control of the agency’s right to disseminate the information would not be reluctant to provide the information. . We need not decide whether the four junior RING officers’ names are protected from disclosure under Exemption 7(C).

The four anonymous letters, which are exempt from disclosure in their entirety under Exemption 7(D), were the only sources of information about these officers.

Furthermore, the allegations against these four officers were not referred to the IG’s Office for investigation, and form no part of the IG report. .Exemption 6, the alternative "privacy” exemption asserted by the Army, protects "personnel and medical files and similar files the disclosure of which would constitute a clearly unwarranted invasion of personal privacy.” (Emphasis added.) Exemption 6 affords the Army far less protection against compelled disclosure than does Exemption 7(C).

Exemption 6 requires proof that the requested documents come within the narrow definition of "similar files,” that the invasion of privacy would be “clearly unwarranted,” and that disclosure would in fact constitute an invasion of privacy. Reporters Comm., 489 U.S. at 756, 109 S.Ct. at 1472; Nadler v. Department of Justice, 955 F.2d 1479, 1488 (11th Cir.1992) (phrase "reasonably expected” represents relaxation of standard from "would constitute," making it easier for agency to invoke Exemption 7(

C) than Exemption 6). For these reasons, we confine our discussion to Exemption 7(C). . Even if the Army had pursued this line of argument on appeal, it is unlikely that the scope of FOIA protection would be broadened. To the extent that the RING officers lacked a sufficient privacy interest in witness statements relating to unsubstantiated, allegations, their privacy interest would surely diminish as to substantiated charges in which there presumably would be a heightened public interest.