OPINION OF THE COURT
TABLE OF CONTENTS
I. Facts and Procedural History...............................................353
A. The Parties and Underlying Causes of Action.............................353
B. The Privilege Dispute..................................................354
II. Jurisdiction...............................................................357
III. Choice of Law.............................................................358
IV. Summary of the Law.......................................................359
A. The Attorney-Client Privilege...........................................359
B. The Disclosure Rule ...................................................361
C. Privileged Information Sharing..........................................362
1. The Co-Client (or Joint-Client) Privilege .............................362
2. The Community-of-Interest (or Common-Interest) Privilege............363
*352D. The Exception for Adverse Litigation....................................366
E. When Joint Representation Goes Awry: The Eureka Principle..............368
F. Putting It All Together: Parents, Subsidiaries, and the Modern Corporate Counsel’s Office............................................369
1. Intra-group Information Sharing: Parents and Subsidiaries as Joint Clients .........................................................369
2. Keeping Control of the Privilege.....................................372
3 When Conflicts Arise...............................................373
V.Issues on Appeal ..........................................................374
A. Whether the Debtors Are Entitled to Documents Generated in the Course of a BCE/Teleglobe Joint Representation........................374
1. Wfliether BCE’s Concession in the Bankruptcy Court Prevents it from Arguing that the Debtors are not Entitled to the Disputed Documents......................................................374
a. Background...................................................374
b. Merits........................................................376
i. Issue Waiver..............................................376
ii. Judicial Admission.........................................377
iii. Judicial Estoppel..........................................377
iv. Implied Prospective Waiver of the Privilege...................378
2. Whether the Community-of-Interest Privilege Entitles the Debtors to the Documents as a Matter of Law...............................378
3. WTiether Teleglobe’s Waiver of the Privilege for the Debtors’ Benefit in the Canadian Insolvency Proceedings Entitles them to the Documents......................................................379
4. Conclusion and Remand ............................................380
B. The Effect of Funneling Documents Through BCE’s In-House Counsel.....380
VI.Potential Alternate Sustaining Grounds.......................................383
A. The Fiduciary Exception to the Attorney-Client Privilege ..................383
B. Affirming as a Discovery Sanction.......................................386
VII.Conclusion................................................................386
This is a twist on a classic corporate divorce story. It begins much as Judge Richard Cudahy’s “classic corporate love story”: “Company A meets Company B. They are attracted to each other and after a brief courtship, they merge.” GSC Partners CDO Fund v. Washington, 368 F.3d 228, 232 (3d Cir.2004). Sadly, it does not last. Not long after Company A acquires Company B, they start taking risks together, some of which go terribly wrong. After only a year or so, Company B is steeped in debt, and, not surprisingly, Company A begins to “los[e] that lovin’ feelin’.”1 It leaves Company B, explaining that it simply must do so in order to save itself. Jilted and out of money, Company B promptly turns to that shelter for abandoned corporations, the bankruptcy system.
In bankruptcy, Company B’s children (subsidiaries), also in the shelter of bankruptcy, become indignant, and they sue Company A for all manner of ills relating to the break-up. Here, we deal not with the merits of the action, but with a pretrial dispute over corporate documents. Everyone agrees that the attorney-client privilege protects these documents against third parties. The wrinkle is that they *353were produced by and in communication with attorneys who represented the entire corporate family back when they all got along.
The question, then, is whether Company A may assert the privilege against its former family members. Because we conclude that the District Court’s factual findings do not support setting aside the parent company’s privilege in this case, we vacate its order compelling production and remand for further proceedings.
1. Facts and Procedural History
A. The Parties and Underlying Causes of Action
This action began with a complaint brought in a Chapter 11 bankruptcy case. The debtors (“Debtors”) are the wholly owned United States subsidiaries of a Canadian telecommunications company formerly known as Teleglobe, Inc. (“Tele-globe”). Teleglobe and the Debtors are undergoing reorganization in Ontario in accordance with the Canadian Companies’ Creditors Arrangement Act (the “Arrangement Act”), a form of bankruptcy protection similar to Chapter 11. In addition, the Debtors (but not Teleglobe), all but one2 of which are Delaware corporations, are simultaneously undergoing Chapter 11 reorganization in the District of Delaware. Until recently, Teleglobe was a wholly owned subsidiary of Bell Canada Enterprises, Inc. (“BCE”), Canada’s largest telecommunications company.3
In 2000, BCE, which had previously owned a 23% minority stake in Teleglobe, purchased all its remaining shares (directly and indirectly through subsidiaries), thus taking control of the company. According to the Debtors, in late 2000 BCE directed Teleglobe to accelerate the development of a fiberoptic network called Glo-beSystem. BCE pledged its financial support to the project and caused Teleglobe and its subsidiaries (the Debtors) to borrow some $2.4 billion from banks and bondholders. The bond debt was guaranteed by one of the Debtors. Teleglobe exhausted its funding in 2001, and in November of that year BCE approved an additional $850 million equity infusion for Teleglobe and its subsidiaries. These monies were to be disbursed at the sole discretion of Jean Monty, then Chairman and CEO of BCE as well as Chairman and CEO of Teleglobe. BCE announced its intention to continue funding Teleglobe in December 2001.
About this time BCE began working on what personnel referred to as Project X— a comprehensive reassessment of BCE’s plans for Teleglobe. Lurking in the background was BCE’s declining confidence in GlobeSystem’s ultimate potential.4 In the course of Project X, BCE considered a variety of options, including maintaining its funding in the hope that GlobeSystem *354would be profitable, restructuring Tele-globe in such a way that it could continue as a viable subsidiary, and simply cutting off funding (which would send Teleglobe and its subsidiaries into a liquidating bankruptcy). In early April 2001, BCE publicly announced that it was reassessing its funding of Teleglobe; just a few weeks later, it ceased its funding, effectively abandoning Teleglobe. GlobeSystem was not operational, and so Teleglobe had no means of paying back its multi-billion dollar debt. Consequently, within weeks Teleglobe and the Debtors filed for Arrangement Act relief in Canada, and the Debtors also filed for Chapter 11 relief in Delaware.
For BCE’s role in funding and then abandoning the GlobeSystem project, the Debtors sued it in this adversary proceeding.5 They assert several causes of action, including breach of contract, breach of fiduciary duties, estoppel, and misrepresentation (whether fraudulent or negligent). All claims relate to the manner in which BCE ceased funding Teleglobe, the Debtors’ corporate parent. Debtors’ theme is that BCE reneged on binding commitments to fund Teleglobe and fraudulently or negligently induced Teleglobe and the Debtors to continue incurring debt in reliance on those commitments, thus harming, inter alia, Teleglobe, the Debtors, and the Debtors’ creditors. Moreover, they allege that BCE, as the controlling shareholder of Teleglobe and the Debtors while those entities were insolvent, breached its fiduciary duties to the Debtors.
B. The Privilege Dispute
In the District of Delaware Bankruptcy Court, the Debtors and the Creditors Committee began exploring through Rule 20046 discovery the possibility of suing BCE for the manner in which it abandoned Teleglobe and the Debtors. In response to discovery requests, BCE marked 98 documents as protected by a “common interest privilege.”7 When the creditors moved to compel production, BCE responded that the documents were privileged because “BCE attorneys consulted with attorneys, officers, or employees of Teleglobe, Inc. or its subsidiaries to discuss or provide legal advice in matters where BCE and Teleglobe, Inc. (or its subsidiaries) shared a common legal interest.” App. at A01110. BCE further stated that the “privilege will continue to exist until the Debtors file a litigation against BCE.” Id.
At a hearing in the Bankruptcy Court, BCE — in keeping with the theme of its argument — agreed to produce (even without the filing of a suit) the “common interest” documents to the Debtors, seemingly agreeing that they fell within the scope of their shared interest, and the Bankruptcy Court entered an order to that effect. BCE did not specifically admit that it was required to produce the documents; it merely agreed to do so. It continued to maintain, however, that the rest of the documents designated as privileged represented advice provided solely to it and were not part of any joint representation with Teleglobe or the Debtors. It is un*355clear from the record exactly what the 98 “common interest” documents contained that BCE agreed to produce, but the privilege logs reflect that they primarily consisted of documents created by BCE’s in-house counsel on the subject of Teleglobe’s financing and restructuring. At the same time, the privilege log (exclusive of the 98 “common interest” documents) lists many other documents reflecting legal advice on Project X matters that BCE claimed— then and now — were intended as advice solely to it and not as part of any joint representation. See generally App. at A00967-A01091.
Once the Debtors and Creditors’ Committee filed their suit against BCE, the District Court withdrew its automatic reference to the Bankruptcy Court and began handling the suit itself. The District Court held an initial discovery conference at which BCE reasserted that it had produced all of the documents that it thought were generated as a result of a BCE/Teleglobe/Debtors joint representation and that the documents it was withholding reflected advice provided to and intended solely for BCE. App. at A0026465. The Debtors did not press the joint representation/common interest point at that time, nor did they argue for a broader scope of the joint representation/common interest than BCE had admitted; rather, they focused on an extension of the “conflicted fiduciary” line of cases, see Part VI.A, infra.8 App. at A00262-63.
The District Court ended up referring the discovery dispute to a Special Master—C.J. Seitz, Jr., Esquire. In its initial written response to the Debtors’ motion to compel production before the Special Master, BCE stated that it had, in response to the Bankruptcy Court’s Rule 2004 order, “produced to the Debtors- all the documents that were protected by a common interest.” App. at A0197. BCE further stated that it had reviewed all of the documents on the privilege log and that the only documents that remained designated as privileged were those “reflecting] the provision of legal work solely to BCE.” Id.
The Debtors then expanded their argument by contending that the scope of the joint representation was broader than BCE admitted. Specifically, they claimed that various attorneys- represented all of the entities on the matters of BCE’s decision to cease funding Teleglobe and Tele-globe’s resulting restructuring. BCE, on the other hand, claimed that it retained its own attorneys to advise it on those matters. The Special Master found that the Debtors had not met their burden of proving an exception to the attorney-client privilege. The evidence, he concluded initially, merely showed that BCE’s in-house counsel represented Teleglobe and the Debtors occasionally; it did not show a broad joint representation as to the abandonment of Teleglobe. Recognizing, however, that the Debtors did not trust BCE’s representation that the documents marked as privileged reflected advice provided solely to BCE, the Special Master ordered a 50-document audit by his in camera *356review to ensure the accuracy of BCE’s representations.
The Special Master also rejected the Debtors’ conflicted fiduciary argument, noting that the Delaware Court of Chancery has refused to adopt it in its broader form. Deutsch v. Cogan, 580 A.2d 100, 105 (Del.Ch.1990) (“Although neither the Garner nor Valente case is binding on this Court, Delaware courts have consistently followed Gamer and declined to broadly apply Valente.”) (citations omitted). More importantly, the Special Master forestalled any further “conflicted fiduciary”-style argument by ruling that neither Teleglobe nor the Debtors’ boards were conflicted in any sense because all of their duties flowed back up to BCE (and not, as the Debtors argued, to their creditors). See Anadarko Petroleum Corp. v. Panhandle Eastern Corp., 545 A.2d 1171, 1174 (Del.1988) (“[I]n a parent and wholly-owned subsidiary context, the directors of the subsidiary are obligated only to manage the affairs of the subsidiary in the best interests of the parent and its shareholders.”). The Special Master, however, did not make an express finding of fact on when the Debtors became insolvent (or entered the amorphous 9 “zone of insolvency”), reasoning instead that there was no way to get from the Debtors’ directors owing fiduciary duties to creditors of the Debtors on one hand to BCE owing a duty to those creditors on the other.
According to the Special Master, “[sjhortly after the Debtors made their selection of documents for in camera review, the wheels started coming off [BCE’s] privilege wagon.” App. at A0030. Before the review, BCE withdrew its privilege assertion for six of the 50 documents that the Debtors selected. Then, the Special Master determined that three of the documents did not involve the provision of legal advice at all, and three lent credence to the Debtors’ argument that BCE attorneys jointly represented BCE and Tele-globe on the issue of BCE’s abandonment. After the initial in camera review, BCE withdrew the privilege assertion for still more documents.
Having reviewed 44 documents in camera, the Special Master issued a supplemental decision in which he concluded that BCE’s revised privilege claims and his review of documents in camera “raised serious questions about” the reliability of the privilege log, whether BCE attorneys jointly represented BCE and Teleglobe on the abandonment issue, whether the documents withheld reflected legal advice provided solely to BCE, and whether the documents withheld were in fact privileged. App. at A0031. He ordered BCE to review and revise the privilege log and to submit all purportedly privileged documents to him for in camera review.
BCE culled its privilege log to just over 1,000 documents. Then, between the submission of the revised privilege log and the submission of the actual documents, BCE withdrew the assertion of privilege for over 100 additional documents. BCE still wasn’t finished; while the in camera review proceeded, it withdrew its assertion of privilege in four separate letters to the Special Master, covering well over 100 more documents.
One of the issues raised by the Debtors in supplemental briefing was BCE’s appar*357ent over-designation of privileged documents. According to the Debtors, this was substantial enough to merit wholesale disclosure of the documents on BCE’s privilege log as a discovery sanction. The Special Master agreed that BCE “failed the audit in multiple ways — -withdrawing documents before in camera review, claiming privilege over documents that did not reflect legal advice, and claiming privilege over documents where it appeared that BCE in-house attorneys and outside counsel jointly represented BCE, Teleglobe, or the Debtors on matters of common interest.”
In his final decision, the Special Master declined to impose disclosure as a discovery sanction. But he nonetheless reversed himself and ordered the production of all of the documents on the privilege log. Having reviewed some 800 documents in camera, he found that they “revealed a broad legal representation of both BCE and Teleglobe by BCE’s in-house attorneys relating to Teleglobe’s restructuring alternatives.” App. at A0047-48. He further found that all of the documents on the privilege log were disclosed to BCE’s in-house counsel, which made them discoverable because those attorneys were jointly representing Teleglobe and could not, therefore, withhold the documents from it. Id. at A0055. He applied this reasoning even to documents produced by outside counsel hired only to work for BCE.10
The District Court affirmed the Special Master’s decision and ordered BCE to turn over to the Debtors all of the documents. BCE argued that the Special Master’s finding of a broad joint representation between it and Teleglobe was irrelevant because he had not found a joint representation between it and the Debtors. Unless the Debtors were a party to the joint representation, BCE argued, they could not invade its privilege. The Court rejected this argument on three grounds: (1) BCE made a binding agreement to disclose all communications generated as part of a BCE/Teleglobe joint representation, and so finding a joint representation between the two was all that was needed; (2) the Debtors, as wholly owned subsidiaries of Teleglobe, were parties to the joint representation as a matter of law; and (3) even the documents that fell outside of the joint representation (ie., were produced by outside counsel) must be disclosed because they were shared with BCE’s in-house attorneys, who jointly represented Teleglobe.
II. Jurisdiction
This is an appeal from an interlocutory order. Nevertheless, we have jurisdiction under the collateral order doctrine, which provides an exception to the finality requirement of 28 U.S.C. § 1291. Under it, an immediate appeal lies if the following elements are met: “(1) the order from which the appellant appeals conclusively determines the disputed question; (2) the order resolves an important issue that is completely separate from the merits of the dispute; and (3) the order is effectively unreviewable on appeal from a final judgment.” In re Ford Motor Co., 110 F.3d 954, 958 (3d Cir.1997) (citing Rhone-Poulenc Rorer, Inc. v. Home Indem. Co., 32 F.3d 851, 860 (3d Cir.1994)).
Here, the first and third prongs are clearly met, as the District Court ordered the production of some 800 documents currently in dispute. See In re Ford Motor Co., 110 F.3d at 958 (holding that the first *358prong is met when a district court orders the production of documents). Once documents are disclosed, any dispute over their privileged status is effectively moot and unreviewable, as the very purpose of the privilege is frustrated by compelled disclosure. Id. at 963 (“[T]he limited assurance that the protected material will not be disclosed at trial ‘will not suffice to ensure free and full communication by clients who do not rate highly a privilege that is operative only at the time of trial.’ ” (quoting Chase Manhattan Bank, N.A. v. Turner & Newall, PLC, 964 F.2d 159, 165 (2d Cir. 1992))).
The only question, then, is whether the privilege issue is sufficiently separate from the merits of the suit. It is, as we have no occasion to consider the issues that lie at the heart of the case: the existence, content, and alleged breach of any contract between BCE, Teleglobe, and/or the Debtors; the content and fraudulent or negligent nature of any representations made by BCE; and the alleged breach by BCE of fiduciary duties it purportedly owed to the Debtors. Rather, we concern ourselves with the extent to which BCE, Tele-globe, and the Debtors were jointly represented by counsel and the effect any joint representation has on BCE’s ability to shield documents from disclosure. In this context, we have little trouble concluding that we have jurisdiction over this appeal.
III. Choice of Law
BCE argues that Canadian law should govern all aspects of this appeal; the Debtors, on the other hand, argue for Delaware law. As a federal court exercising jurisdiction over state-law claims, we apply the choice-of-law rules of Delaware, the forum state. Hammersmith v. TIG Ins. Co., 480 F.3d 220, 226 (3d Cir.2007). “Delaware courts look to the Restatement (Second) of Conflict of Laws for guidance in choice of law disputes.” Gloucester Holding Corp. v. U.S. Tape & Sticky Prods., LLC, 832 A.2d 116, 124 (Del.Ch. 2003). Here, § 139 of the Restatement applies:
(1) Evidence that is not privileged under the local law of the state which has the most significant relationship with the communication will be admitted, even though it would be privileged under the local law of the forum, unless the admission of such evidence would be contrary to the strong public policy of the forum.
(2) Evidence that is privileged under the local law of the state which has the most significant relationship with the communication but which is not privileged under the local law of the forum will be admitted unless there is some special reason why the forum policy favoring admission should not be given effect.
Restatement (Second) of Conflict of Laws § 139 (1971).
These provisions presuppose a conflict between the law of the forum state and the law of the state with the most significant relationship. This is because courts typically wade into choice-of-law determinations when those laws truly conflict. See Huber v. Taylor, 469 F.3d 67, 74 (3d Cir.2006) (citing On Air Entm’t Corp. v. Nat’l Indem. Co., 210 F.3d 146, 149 (3d Cir.2000)). While there are no reported Delaware cases on this point, we predict that Delaware would follow the practice of the federal system and most states, and decide a choice-of-law dispute only when the proffered legal regimes actually conflict on a relevant point.
BCE argues for Canadian law here, but concedes that it has found no relevant conflict of Canadian with Delaware law.11 *359This undercuts BCE’s argument, particularly given that the Restatement favors the admission of evidence when the law of the forum state so requires; thus, if applying Canadian law will safeguard the privilege, it is BCE’s responsibility not only to highlight the legal conflict, but also to point to “some special reason” favoring protection. See Restatement (Second) of Conflict of Laws § 139 emts. c & d. As BCE has not informed us of a conflict, cites Delaware law extensively (with comparatively few Canadian decisions noted), and concedes that the law in Canada is not materially different than Delaware, we rely on Delaware authority in reaching our conclusions. See Huber, 469 F.3d at 74.
We recognize that BCE has informed us of a true conflict on an issue the Debtors raise as an alternate sustaining ground— namely, whether the Debtors are entitled to the disputed documents under the fiduciary exception to the attorney-client privilege. Though that exception is well-entrenched in Delaware law, it does not exist in Canada. We deal with this issue more fully in Part VI.A.
IV. Summary of the Law
This appeal raises core questions about the proper operation of a corporate family’s centralized in-house legal department. Because answering those questions requires delving into a variety of concepts related to the co-client (or joint-client) privilege, its exceptions, its scope, and a lawyer’s ethical obligations, a summary of relevant law sets the backdrop.
We begin with a basic review of the attorney-client privilege — and how it has been adapted to accommodate the corporate client. We continue with a discussion of how disclosing otherwise privileged communications to third parties waives the privilege. Next, we explore two oft-confused privileges: (1) the co-client (or joint-client) privilege, which applies when multiple clients hire the same counsel to represent them on a matter of common interest,12 and (2) the community-of-interest (or common-interest) privilege, which comes into play when clients with separate attorneys share otherwise privileged information in order to coordinate their legal activities. Neither the co-client nor community-of-interest privilege is effective in adverse litigation between the former clients, so we next discuss the contours of the adverse-litigation exception. Then, we explore how courts deal with joint representations that go wrong because of impermissible attorney conflicts of interest. Finally, we put all of these doctrines together to address how they interact in the modern corporate in-house counsel’s office.
A. The Attorney-Client Privilege
The attorney-client privilege protects communications between attorneys and clients from compelled disclosure. It applies to any communication that satisfies the following elements: it must be “(1) a communication (2) made between privileged persons (3) in confidence (4) for the purpose of obtaining or providing legal assistance for the client.” Restatement (Third) of the Law Governing Lawyers § 68 (2000). “Privileged persons” include the client, the attorney(s), and any of their agents that help facilitate attorney-client communications or the legal representation. Id. § 70.
The attorney-client privilege is the oldest of the common-law privileges. Klitzman, Klitzman & Gallagher v. Krut, 744 F.2d 955, 960 (3d Cir.1984). Like all privileges, it is an exception to the common-law maxim that the public has a right to “ ‘ev*360ery man’s evidence.’ ” United States v. Bryan, 339 U.S. 323, 331, 70 S.Ct. 724, 94 L.Ed. 884 (1950) (quoting 8 J. Wigmore, Evidence § 2192 (3d ed.1940)). Though initially confined to communications made in anticipation of litigation, American courts rejected that limitation at the outset. Paul R. Rice, Attorney-Client Privilege in the United States § 1:12 (2d ed.1999) (hereinafter “Rice”) (citing, e.g., Parker v. Carter, 18 Va. 273, 4 Munf. 273, 1814 WL 667, at *9 (1814) (“The court is also of [the] opinion[ ] that [the privilege] is not confined to facts disclosed, in relation to suits actually depending at the time, but extends to all cases in which a client applies ... to his counsel or attorney ... for his aid in the line of his profession.”)). This is because so confining the privilege would discourage clients from seeking the advice of counsel before problems arise. Parker, 1814 WL 667, at *9.
As the Supreme Court has noted more recently, the purpose of the attorney-client privilege
is to encourage full and frank communication between attorneys and their clients and thereby promote broader public interests in the observance of law and administration of justice. The privilege recognizes that sound,legal advice or advocacy serves public ends and that such advice or advocacy depends upon the lawyer’s being fully informed by the client.
Upjohn Co. v. United States, 449 U.S. 383, 389, 101 S.Ct. 677, 66 L.Ed.2d 584 (1981); accord The Queen v. McClure, [2001] 1 S.C.R. 445, at ¶¶ 32-33 (Can.2001); Deutsch v. Cogan, 580 A.2d at 104. Because the privilege carries through policy purposes — encouraging attorney-client communication to enhance compliance -with the law and facilitating the administration of justice, the Supreme Court has not applied it mechanically. Upjohn, 449 U.S. at 392, 101 S.Ct. 677. It is essential that parties be able to determine in advance with a high degree of certainty whether communications will be protected by the privilege. Id. at 393, 101 S.Ct. 677; see also Deutsch, 580 A.2d at 106.
As common-law courts developed the privilege in an age in which clients were almost exclusively natural persons, more modern courts sought to adapt it to the now ubiquitous corporate client. For more than a century, common-law courts have recognized that communications between corporate clients and their attorneys are indeed privileged. See Radiant Burners, Inc. v. Am. Gas Ass’n, 320 F.2d 314, 319 & n. 7 (7th Cir.1963) (citing English cases dated as early as 1833, and American cases as early as 1885, for the proposition that corporations can assert the attorney-client privilege).
Because corporations act through human agents, the question of whose communications with the corporation’s attorneys are entitled to protection comes up often. Rice § 4:11. In Upjohn the Supreme Court rejected the so-called “control group theory” — that only communications between high-level managers and corporate attorneys merit protection. 449 U.S. at 392, 101 S.Ct. 677. Instead, it held that when a corporation’s managers require its employees to give information to its attorneys in the course of providing legal advice, those communications also are protected. Id. at 396, 101 S.Ct. 677. This serves the policy goals of the privilege — to enhance compliance with the law and facilitate the administration of justice — by encouraging open communication between attorneys and clients. Id. at 389, 394, 101 S.Ct. 677.
The lesson for us is that it is important not to confuse these overarching policy goals with the means of achieving them. Communication between counsel and client *361is not, in and of itself, the purpose of the privilege; rather, it only protects the free flow of information because it promotes compliance with law and aids administration of the judicial system. Cf. United States v. Zolin, 491 U.S. 554, 563, 109 S.Ct. 2619, 105 L.Ed.2d 469 (1989) (explaining that attorney-client communication facilitating fraud is not privileged because that sort of communication impedes the administration of justice). Thus, following Upjohn’s lead in not applying the privilege mechanically does not counsel in favor of applying the privilege anytime it might increase the flow of information; rather, Upjohn counsels a more nuanced inquiry into whether according a type of communication protection is likely to encourage compliance-enhancing communication that makes our system for resolving disputes more operable.
B. The Disclosure Rule
A communication is only privileged if it is made “in confidence.” Restatement (Third) of the Law Governing Lawyers § 68. In other words, if persons other than the client, its attorney, or their agents are present, the communication is not made in confidence, and the privilege does not attach. The disclosure rule operates as a corollary to this principle: if a client subsequently shares a privileged communication with a third party, then it is no longer confidential, and the privilege ceases to protect it. See Del. R. Evid. 510. This is because the act of disclosing signals that the client does not intend to keep the communication secret. Rice § 9:28. In addition, it prevents clients from engaging in strategic selective disclosure. United States v. Bernard, 877 F.2d 1463, 1465 (10th Cir.1989). The privilege does, after all, hinder the truth-seeking process, and so we carefully police its use. United States v. Doe, 429 F.3d 450, 453 (3d Cir. 2005).13
Disclosing a communication to a third party unquestionably waives the privilege. A harder question is whether the waiver also ends the privilege as to any related but not disclosed communications. In answering this question, our touchstone is fairness. See Tackett v. State Farm Fire & Cas. Ins. Co., 653 A.2d 254, 259 (Del.1995); see also Westinghouse Elec. Corp. v. Republic of the Philippines, 951 F.2d 1414, 1426 n. 12 (3d Cir.1991). When one party takes advantage of another by selectively disclosing otherwise privileged communications, courts broaden the waiver as necessary to eliminate the advantage. Zirn v. VLI Corp., 621 A.2d 773, 781-82 (Del.1993) (“The purpose underlying the rule of partial disclosure is one of fairness to discourage the use of the privilege as a litigation weapon.”); see also Rice § 9:31. Extending the waiver, however, is not a punitive measure, so courts do not imply a broader waiver than necessary to ensure that all parties are treated fairly.14 See Rice 9:31. Moreover, when the disclosure does not create an unfair advantage, courts typically limit the waiver to the communications actually disclosed. See In re Keeper of Records (Grand Jury Subpoena Addressed to XYZ Corp.), 348 F.3d 16, 24-25 (1st Cir.2003); cf. Westinghouse, 951 F.2d at 1427 n. 14.
*362C. Privileged Information Sharing
1. The Co-Client (or Joint-Client) Privilege
It is often expedient for two or more people to consult a single attorney. The rules of professional conduct allow a lawyer to serve multiple clients on the same matter so long as all clients consent, and there is no substantial risk of the lawyer being unable to fulfill her duties to them all. Restatement (Third) of the Law Governing Lawyers §§ 128-31. Just as in single-client representation, the lawyer and co-clients15 begin their relationship when the co-clients convey their desire for representation, and the lawyer consents. Id. § 14. Like single-client representation, nothing prevents joint representation from arising by implication, but, as a District Court in Maryland recently noted, courts must be careful not to imply joint representations too readily:
What the Court takes exception to is [the plaintiffs] effort to ... argue, in effect, that a joint representation of Party A and Party B may somehow arise through the expectations of Party B alone, despite Party A’s views to the contrary. This position is untenable, because it would ... allow the mistaken (albeit reasonable) belief by one party that it was represented by an attorney ... to serve to infiltrate the protections and privileges afforded to another client.
Neighborhood Dev. Collaborative v. Murphy, 233 F.R.D. 436, 441-42 (D.Md.2005) (internal citations and quotation marks deleted). Moreover, as the Restatement details, it is important to remember that
clients of the same lawyer who share a common interest are not necessarily co-clients. Whether individuals have jointly consulted a lawyer or have merely entered concurrent but separate representations is determined by the understanding of the parties and the lawyer in light of the circumstances.
Co-client representations must also be distinguished from situations in which a lawyer represents a single client, but another person with allied interests cooperates with the client and the client’s lawyer.
Restatement (Third) of the Law Governing Lawyers § 75 cmt. c (internal cross-references omitted).
Once begun, a co-client representation generally continues until a client discharges the lawyer or the lawyer withdraws. Id. § 31. In addition, numerous courts have recognized that the relationship may terminate by implication. Fed. Deposit Ins. Corp. v. Ogden Corp., 202 F.3d 454, 463 (1st Cir.2000) (“A joint attorney-client relationship remains intact until it is expressly terminated or until circumstances arise that readily imply to all the joint clients that the relationship is over.”) (emphasis in original); see also Flynt v. Brownfield, Bowen, & Bally, 882 F.2d 1048, 1051 (6th Cir.1989) (holding that terminating attorney-client relationship requires “conduct dissolving the essential mutual confidence” between them); In re Dow, 132 B.R. 853, 858 (Bankr.S.D.Ohio 1991) (same). In particular, a joint representation terminates when it becomes clear to all parties that the clients’ legal interests have diverged too much to justify using common attorneys. Rice § 2:4 (citing Mass. Eye & Ear Infirmary v. QLT Phototherapeutics, Inc., 167 F.Supp.2d 128, 129 (D.Mass.2001)).
While it is permissible for lawyers and clients to limit the scope of representation in a single-client representation, see Restatement (Third) of the Law Govern*363ing Lawyers § 19, it is particularly common in co-client situations because of the limited congruence of the clients’ interests. As the Restatement notes, a co-client relationship is limited by “the extent of the legal matter of common interest.” Id. § 75 cmt. c. While written agreements limiting the scope of a joint representation might be preferable, nothing requires this so long as the parties understand the limitations.
The District Court for the Northern District of California noted in Sky Valley Ltd. P’ship v. ATX Sky Valley Ltd., 150 F.R.D. 648, 652-53 (N.D.Cal.1993), that a wide variety of circumstances are relevant to the determination of whether two or more parties intend to create a joint-client relationship, particularly how the parties interact with the joint attorneys and with each other. These same circumstances are relevant to determining the scope of any joint representation.16 The keys to deciding the scope of a joint representation are the parties’ intent and expectations, and so a district court should consider carefully (in addition to the content of the communications themselves) any testimony from the parties and their attorneys on those areas. As explained in section D of this Part, finding too broad the scope of a joint representation gives the parties more control over each other’s ability to waive the privilege than they intended, and it subjects them to losing it in litigation with one another.
When co-clients and their common attorneys communicate with one another, those communications are “in confidence” for privilege purposes. Hence the privilege protects those communications from compelled disclosure to persons outside the joint representation. Moreover, waiving the joint-client privilege requires the consent of all joint clients. Restatement (Third) of the Law Governing Lawyers § 75(2). A wrinkle here is that a client may unilaterally waive the privilege as to its own communications with a joint attorney, so long as those communications concern only the waiving client; it may not, however, unilaterally waive the privilege as to any of the other joint clients’ communications or as to any of its communications that relate to other joint clients.17 Id. at cmt. e.
2. The Community-of-Interest (or Common-Interest) Privilege18
Recognizing that it is often preferable for co-defendants represented by different attorneys in criminal proceedings to coor*364dinate their defense, courts developed the joint-defense privilege. In its original form, it allowed the attorneys of criminal co-defendants to share confidential information about defense strategies without waiving the privilege as against third parties. Moreover, one co-defendant could not waive the privilege that attached to the shared information without the consent of all others.19 Later, courts replaced the joint-defense privilege, which only applied to criminal co-defendants, with a broader one that protects all communications shared within a proper “community of interest,” whether the context be criminal or civil.20 Rice § 4:35; see also Andrew R. Taggart, Parent-Subsidiary Communications & the Attorney-Client Privilege, 65 U. Chi. L.Rev. 315 (1998). Thus, the community-of-interest privilege allows attorneys representing different clients with similar legal interests to share information without having to disclose it to others. It applies in civil and criminal litigation, and even in purely transactional contexts. Rice § 4:35; Restatement (Third) op the Law Governing Lawyers § 76.
Two aspects of the modern community-of-interest privilege are noteworthy. First, to be eligible for continued protection, the communication must be shared with the attorney of the member of the community of interest. Cf. Ramada Inns, Inc. v. Dow Jones & Co., 523 A.2d 968, 972 (Del.Super.Ct.1986) (emphasizing that the relevant Delaware evidentiary rule protects communications disclosed to an attorney). Sharing the communication directly with a member of the community may destroy the privilege.21 Second, all members of the community must share a common legal interest in the shared communication. Rice § 4:35. Delaware Rule of Evidence 502(b)(3), which sets out the State’s version of the community-of-interest privilege, incorporates both requirements (that the clients’ separate attorneys share information and that the clients have a common legal interest):
A client has a privilege to refuse to disclose and to prevent any other person from disclosing confidential communications[,] made for the purpose of facilitating the rendition of professional legal services to the client ... [,] by the client or the client’s representative or the client’s lawyer or a representative of the lawyer to a lawyer or a representative of a lawyer representing another in a matter of common interest.
Del. R. Evid. 502(b)(3).
The requirement that the clients’ separate attorneys share information (and not the clients themselves) derives from the community-of-interest privilege’s roots in the old joint-defense privilege, which (to repeat) was developed to allow attorneys to coordinate their *365clients’ criminal defense strategies. See Chahoon v. Commw., 62 Va. 822, 21 Gratt. 822, 1871 WL 4931, at *11 (1871). Because the common-interest privilege is an exception to the disclosure rule, which exists to prevent abuse, the privilege should not be used as a post hoc justification for a client’s impermissible disclosures. The attorney-sharing requirement helps prevent abuse by ensuring that the common-interest privilege only supplants the disclosure rule when attorneys, not clients, decide to share information in order to coordinate legal strategies.
Similarly, the congruence-of-legal-interests requirement ensures that the privilege is not misused to permit unnecessary information sharing. In a leading case, a District Court in South Carolina explained the contours of the requirement:
A community of interest exists among different persons or separate corporations where they have an identical legal interest with respect to the subject matter of a communication between an attorney and a client concerning legal advice. The third parties receiving copies of the communication and claiming a community of interest may be distinct legal entities from the client receiving the legal advice and may be a non-party to any anticipated or pending litigation. The key consideration is that the nature of the interest be identical, not similar, and be legal, not solely commercial. The fact that there may be an overlap of a commercial and a legal interest for a third party does not negate the effect of the legal interest in establishing a community of interest.
Duplan Corp. v. Deering Milliken, Inc., 397 F.Supp. 1146, 1172 (D.S.C.1974).
The Restatement takes a more flexible approach than Duplan toward the similarity and types of interests that qualify as “common”: “[T]he common interest ... may be either legal, factual, or strategic in character. The interests of the separately represented clients need not be entirely congruent.” Restatement (Third) of the Law Governing Lawyers § 76 cmt. e. Professor Rice criticizes Duplan’s strictness and cites a few cases announcing that the interest need not be identical (though maintaining, contrary to the Restatement, that the interest must be legal, rather than “factual or strategic”). Rice § 4:36 (citing, e.g., SCM Corp. v. Xerox Corp., 70 F.R.D. 508, 524-25 (D.Conn.1976) (Newman, J.) (holding that legal interests must be “demonstrably common” or the clients must have a “substantial” risk of shared exposure to justify privileged information-sharing)). Rice, however, still recognizes Duplan as the leading approach. Id. (“This ... standard ... coined in Duplan has been widely followed.”). The Delaware courts seem not to have taken a position on whether the common legal interest must be identical, and we need not resolve the congruenee-of-legal-interests question here. For our purposes, it is sufficient to recognize that members of the community of interest must share at least a substantially similar legal interest.
We conclude with two points of caution. First, the privilege only applies when clients are represented by separate counsel. Thus, it is largely inapplicable to disputes like this one that revolve around corporate family members’ use of common attorneys (namely, centralized in-house counsel).22 Second, while the Restatement *366(confusingly) uses the term “common interest” to describe the congruence of the parties’ interests in both co-client and community-of-interest situations, the concepts are not the same. Compare Restatement (Third) of the Law Governing Lawyers § 75(1) (“If two or more persons are jointly represented by the same lawyer in a matter, a communication of either co-client that ... relates to matters of common interest is privileged as against.third persons.”), with id. § 76(1) (“If two or more clients with a common interest in a litigated or nonlitigated matter are represented by separate lawyers and they agree to exchange information concerning the matter, a communication of any such client ... is privileged as against third persons.”); cf. id. § 76 cmt. e & Reporter’s Note cmt. b (explaining that co-client and community-of-interest situations differ). In particular, because co-clients agree to share all information related to the matter of common interest with each other and to employ the same attorney, their legal interests must be identical (or nearly so) in order that an attorney can represent them all with the candor, vigor, and loyalty that our ethics require. See Ogden, 202 F.3d at 461. In the community-of-interest context, on the other hand, because the clients have separate attorneys, courts can afford to relax the degree to which clients’ interests must converge without worrying that their attorneys’ ability to represent them zealously and single-mindedly will suffer.
D. The Exception for Adverse Litigation
The great caveat of the joint-client privilege is that it only protects communications from compelled disclosure to parties outside the joint representation. When former co-clients sue one another, the default rule is that all communications made in the course of the joint representation are discoverable. Del. R. Evid. § 502(d)(6); Restatement (Third) of the Law Governing Lawyers § 75(2). This rule has two bases: (1) the presumed intent of the parties, and (2) the lawyer’s fiduciary obligation of candor to both parties. Id. § 75 cmt. d. According to the Restatement, it is permissible for co-clients to agree in advance to shield information from one another in subsequent adverse litigation, though the drafters concede finding no direct authority for that proposition.23 Id. Reporter’s Note cmt d.
BCE argues that the default rule should be flipped when the joint clients are a parent company and its wholly owned subsidiary — i.e., courts should assume that communications generated in the course of the joint representation are not discoverable in adverse litigation. BCE’s rationale is that no parent would want its subsidiary to be able to invade the privilege in subsequent litigation, and so courts should not presume that intent. Moreover, because parents and wholly owned subsidiaries share the same interests, the parent’s intent (to shield information) effectively controls.
Delaware courts have recognized that parents and their wholly owned subsidiaries have the same interests because all of the duties owed to the subsidiaries flow back up to the parent. Cf. Anadarko, 545 A.2d at 1174 (“[I]n a parent and wholly-owned subsidiary context, the directors of the subsidiary are obligated *367only to manage the affairs of the subsidiary in the best interests of the parent and its shareholders”). While we normally assume that a corporation’s primary interest is in maximizing its economic value, the only interest of a wholly owned subsidiary is in serving its parent. Id. at 1174. That doing so may not always involve maximizing the subsidiary’s economic value is of little concern. Trenwick Am. Litig. Trust v. Ernst & Young, L.L.P., 906 A.2d 168, 192 (Del.Ch.2006).24 If the subsidiary is not wholly owned, however, in the interest of protecting minority shareholders we revert to requiring that whoever controls the subsidiary seek to maximize its economic value with requisite care and loyalty. See id. at 192 n. 66. Similarly, if the subsidiary is insolvent, we require the same in the interest of protecting the subsidiary’s creditors. Id. at 204 n. 96 (approving In re Scott Acquisition Corp., 344 B.R. 283, 286 (Bankr.D.Del.2006) (holding that the fiduciary duties of directors of an insolvent wholly owned subsidiary inure to the benefit of the subsidiary’s creditors)); see also N. Am. Catholic Programming Found., 930 A.2d at 101-02, 2007 WL 1453705, at *1 (“[T]he creditors of an insolvent corporation have standing to maintain derivative claims against directors on behalf of the corporation for breaches of fiduciary duties. The corporation’s insolvency makes the creditors the principal constitu-eney injured by any fiduciary breaches that diminish the firm’s value. Therefore, equitable considerations give creditors standing to pursue derivative claims against the directors of an insolvent corporation. Individual creditors of an insolvent corporation have the same incentive to pursue valid derivative claims on its behalf that shareholders have when the corporation is solvent.”) (emphasis in original) (internal citations and quotation marks omitted)).
In the context of a joint attorney representing a parent and its solvent wholly owned subsidiary, BCE’s argument that we should flip the normal default rule (that all information shared in the course of a joint representation is not privileged in subsequent adverse litigation between the former joint clients) has some appeal, as it probably is more in line with the typical parent company’s intent. But because parent-subsidiary relationships often change, having opposite default rules for wholly owned, solvent subsidiaries, and not-wholly owned or insolvent subsidiaries, seems unwieldy. In the course of a joint representation, a subsidiary could go from being wholly owned and solvent to majority-owned or insolvent (or both). Under those circumstances, it is not clear which default rule BCE would have us apply. Because of the need for clarity and cer*368tainty in privilege law, see Upjohn, 449 U.S. at 393, 101 S.Ct. 677, creating multiple, ever-shifting default rules would be unwise. Simply following the default rule against information shielding creates simpler, and more predictable, ground rules.
Moreover, BCE’s argument overlooks the joint attorney’s fiduciary obligations to both parties. In undertaking a joint representation, the prospective joint attorney must always consider whether she can fulfill her duties to each co-client. See Restatement (Third) of the Law Governing Lawyers §§ 128-31. When the co-clients desire to shield information from one another, this inquiry becomes more difficult in light of the prospective joint attorney’s duty of candor to each. See id. § 60 cmt. 1. In situations in which the subsidiary may become either partially owned or insolvent, a prospective joint attorney would have to consider her ability not only to counsel both clients, but to do so in the face of an information-shielding agreement that could harm one of them. Thus applying the default rule against information shielding, absent an affirmative agreement to shield, makes good sense. The rule is widely accepted and dovetails with the joint attorney’s duty of candor.
We predict that Delaware courts would apply the adverse litigation exception in all situations, even those in which the joint clients are wholly owned by the same person or entity.
E. When Joint Representation Goes Awry: The Eureka Principle
The Restatement’s conflicts rules provide that when a joint attorney sees the co-clients’ interests diverging to an unacceptable degree, the proper course is to end the joint representation. Restatement (Third) of the Law Governing Lawyers § 121 cmts. e(1)-(2). As the Court of Appeals for the D.C. Circuit noted in Eureka Inv. Corp. v. Chicago Title Ins. Co., 743 F.2d 932 (D.C.Cir.1984) (per curiam), courts are presented with a difficult problem when a joint attorney fails to do that and instead continues representing both clients when their interests become adverse. Id. at 937-38. In this situation, the black-letter law is that when an attorney (improperly) represents two clients whose interests are adverse, the communications are privileged against each other notwithstanding the lawyer’s misconduct. Id.; see also 8 J. Wigmore, Evidence § 2312 (McNaughton rev. ed.1961).
The context of the Eureka case was the joint representation of an insured and insurer. Over the course of the litigation, the parties began to disagree. The insured, a developer trying to effect a condo conversion, wanted to settle in order to get its conversion plans back on track. The title insurer, on the other hand, wanted to continue opposing liability and would not agree to within-poliey-limits settlement terms. Having decided that the insurer’s refusal to settle was tortious, the insured entered into a unilateral settlement with its adversaries and promptly sued the insurer for indemnification and consequential damages. The trouble was that the insured continued to use the joint attorneys throughout the process, relying on their advice in deciding to enter into a unilateral settlement and to sue the insurer. Thus, upon filing its action, the insurer sought discovery of the insured’s communications with the joint attorneys in the hope that those communications would support the affirmative defense of noncooperation. The insurer argued that because those communications were generated during the attorneys’ joint representation of the parties on the claim against the insured, they were discoverable in an action between the joint clients.
The Court rejected the insurer’s argument, holding instead that “[t]he policy *369behind [the co-client privilege] — to encourage openness and cooperation between joint clients — does not apply to matters known at the time of communication not to be in the common interest of the attorney’s two clients.” Eureka, 743 F.2d at 937. It emphasized that both the insured and the joint attorneys thought that they had begun a separate, individual representation of the insured on the insurance bad-faith claim that was distinct from the underlying liability action, calling these understandings “crucial.” Id. Noting the attorneys’ potential ethical violations, the Court concluded that they were of no moment: “[C]ounsel’s failure to avoid a conflict of interest should not deprive the client of the privilege. The privilege, being the client’s, should not be defeated solely because the attorney’s conduct was ethically questionable.” Id. at 938. Though not yet explicitly adopted by the Delaware courts, the Eureka principle is widely accepted. See Restatement (Third) of the Law GovERNING LAWYERS § 60 cmt. 1.; Rice § 4:33.
F. Putting It All Together: Parents, Subsidiaries, and the Modern Corporate Counsel’s Office
“A striking development in the legal profession ... has been the rapid growth in both importance and size of in-house, or corporate, counsel.” Abram Chayes & Antonia Chayes, Corporate Counsel and the Elite Law Firm, 37 Stan. L.Rev. 277, 277 (1985). The roles of in-house counsel are many, e.g., overseeing the corporation’s compliance with myriad regulatory regimes. The primary advantages of in-house (rather than outside) counsel are the breadth of their knowledge of the corporation and their ability to begin advising senior management on important transactions at the earliest possible stage, often well before anyone would think to hire a law firm. See id. at 280-81; Carl D. Liggio, The Changing Role of Corporate Counsel, 46 Emory L.J. 1201, 1208 (1997). While there is much debate over how corporate counsel should go about promoting compliance with law (e.g., the usefulness of “noisy withdrawal” requirements versus going up the corporate chain with concerns), both sides of the debate seem to see in-house counsel as the “front lines” of the battle to ensure that compliance while preserving confidential communications. Compare William W. Horton, A Transactional Lawyer’s Perspective on the Attorney-Client Privilege: A Jeremiad for Upjohn, 61 Bus. Law. 95 (2005), with Letter from Susan P. Koniak, Roger C. Cramton & George M. Cohen (endorsed by named academies) to Securities and Exchange Commission (Dec. 17, 2002), available at http://www.sec.gov/rules/proposed/s74502/ skoniakl.htm. Because in-house counsel are crucial and clear rules are needed to sort out attorney-client privilege problems (particularly for corporate groups), this section sets out how the various principles we have discussed apply to a parent company’s in-house counsel.
1. Intra-group Information Sharing: Parents and Subsidiaries as Joint Clients
Because parent companies often centralize the provision of legal services to the entire corporate group in one in-house legal department, it is important to consider how the disclosure rule affects the sharing of information among corporate affiliates. Recognizing that any other result would wreak havoc on corporate counsel offices, courts almost universally hold that intra-group information sharing does not implicate the disclosure rule. This result is unquestionably correct. The cases, however, vary in how they reach the result. The Glidden case, which both parties cite, illustrates the conceptual muddle:
The universal rule of law, expressed in a variety of contexts, is that the parent and subsidiary share a community of *370interest, such that the parent (as well as the subsidiary) is the “client” for purposes of the attorney-client privilege. See Crabb v. KFC Nat’l Management Co., No. 91-5474, 1992 WL 1321 (6th Cir. Jan.6, 1992) (“ ‘The cases clearly hold that a corporate ‘client’ includes not only the corporation by whom the attorney is employed or retained, but also parent, subsidiary and affiliate corporations.’ ” (quoting United States v. AT & T, 86 F.R.D. 603, 616 (D.D.C.1979))). Consequently, disclosure of legal advice to a parent or affiliated corporation does not work a waiver of the confidentiality of the document, because of the complete community of interest between parent and subsidiary. Id. at *2. Numerous courts have recognized that, for purposes of the attorney client privilege, the subsidiary and the parent are joint clients, each of whom has an interest in the privileged communications. See, e.g., Polycast Tech. Corp. v. Uniroyal, Inc., 125 F.R.D. 47, 49 (S.D.N.Y.1989); Medcom Holding Co. v. Baxter Travenol Lab., 689 F.Supp. 841, 842 (N.D.Ill. 1988).
Glidden Co. v. Jandemoa, 173 F.R.D. 459, 472-73 (W.D.Mich.1997) (applying Delaware law). In the above-quoted paragraph, the Court calls the members of the corporate family a single client and joint clients — all in the same breath. Moreover, it invokes the community-of-interest privilege, which we explained in section C of this Part, supra, applies only when the parties have separate counsel. In quoting Glidden, we do not mean to single out that Court for criticism, for there are many cases that reach the result of non-waiver of privilege without persuasively explaining how.
Courts typically offer versions of three arguments for not construing the sharing of communications within the corporate family as a waiver: (1) the members of the corporate family comprise one client, see, e.g., Glidden, 173 F.R.D. at 472; United States v. Am. Tel. & Tel. Co., 86 F.R.D. 603, 616 (D.D.C.1979); (2) the members of the corporate family are joint clients, see, e.g., Glidden, 173 F.R.D. at 473; Polycast Tech. Corp. v. Uniroyal, Inc., 125 F.R.D. 47, 49 (S.D.N.Y.1989); Medcom, 689 F.Supp. at 842; and (3) the members of the corporate family are in a community of interest with one another. See, e.g., Glidden, 173 F.R.D. at 472; see generally John K. Villa, Corporate Counsel Guidelines § 1:22(C) (2006) (discussing community-of-interest and joint-client rationales); Rice § 4:24 & n. 54 (discussing various rationales for not applying the disclosure rule). Of these three rationales, we believe only the second withstands scrutiny.
Within the wholly owned corporate family, it superficially makes sense to hold, as BCE urges, that the family is really one client for purposes of the privilege and that the privilege is held exclusively by the parent because all fiduciary duties flow to the parent. Indeed, in other contexts courts have treated parents and their wholly owned subsidiaries as one entity. For example, parents and their wholly owned subsidiaries cannot constitute a “combination” or “conspiracy” of two or more persons under the Sherman Act. Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 772, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984). The Supreme Court reasoned that parents and subsidiaries could not effectively “agree” for Sherman Act purposes because
[a] parent and its wholly owned subsidiary have a complete unity of interest. Their objectives are common, not disparate; their general corporate actions are guided or determined not by two separate corporate consciousnesses, but one. They are not unlike a multiple team of horses drawing a vehicle under the control of a single driver. With or without *371a formal “agreement,” the subsidiary acts for the benefit of the parent, its sole shareholder.... [I]n reality a parent and a wholly owned subsidiary always have a unity of purpose or a common design. They share a common purpose whether or not the parent keeps a tight rein over the subsidiary; the parent may assert full control at any moment if the subsidiary fails to act in the parent’s best interests.
Id. at 771-72, 104 S.Ct. 2731 (internal citations and quotation marks omitted). Applying similar logic, courts have held that a parent cannot tortiously interfere with its subsidiary’s contracts merely by directing the subsidiary to breach the agreements. Boulevard Assocs. v. Sovereign Hotels, Inc., 72 F.3d 1029, 1036 (2d Cir.1995) (applying Connecticut law).
BCE also points out that Congress has recently pushed members of corporate families to act more in concert with one another by requiring the officers of public companies to certify each quarter that they “have designed such internal controls to ensure that material information relating to the issuer and its consolidated subsidiaries is made known to such officers by others within those entities.” 15 U.S.C. § 7241(a)(4)(B). According to SEC regulations, the general rule is that majority-owned subsidiaries must be consolidated for purposes of financial reporting. 17 C.F.R. § 210.3A-02(a). Thus our regulatory structure also treats the corporate family as a unified enterprise for at least some purposes.
On the other hand, treating members of a corporate family as one client fails to respect the corporate form. It is a bedrock principle of corporate law in Delaware and elsewhere that courts must respect entity separateness unless doing so would work inordinate inequity. Pauley Petroleum, Inc. v. Cont’l Oil Co., 43 Del. Ch. 516, 239 A.2d 629, 633 (1968); see also In re Owens Corning, 419 F.3d 195, 211 (3d Cir.2005) (concluding that courts must “respect entity separateness absent compelling circumstances calling equity ... into play”). By structuring its various activities by forming separate corporations, a parent company realizes numerous benefits, not the least of which are the liability shields. With that structure comes the responsibility to treat the various corporations as separate entities. In a tort suit, for example, it is doubtful that we would treat the Debtors, Teleglobe, and BCE as one entity.
We acknowledge that a core concept of Copperweld and Boulevard Associates is that when a legislature seeks to attach conspiracy-like liability,25 it rarely means to include the garden-variety situation in which one who controls a corporation directs that corporation to do something. After all, a corporation can only act at the direction of whoever controls it, and we do not think of every corporate action as a “conspiracy.” These decisions, though, are tethered to the statutes (or common-law causes of action) they interpret,26 and do not give us license to disregard entity separateness in other contexts. Thus they are inapplicable here.
Put simply, BCE wants to have it both ways: it wants us to view the corporate group as a single client, and it wants the controlling entity to own the privilege in *372perpetuity. But the Supreme Court held in Commodity Futures Trading Commission v. Weintraub, 471 U.S. 343, 350, 105 S.Ct. 1986, 85 L.Ed.2d 372 (1985), that control of the privilege passes with control of the corporation, so it is unclear (even accepting BCE’s theory) that it is the initial corporate parent who should control the privilege unilaterally once the group breaks up. In any event, absent some compelling reason to disregard entity separateness, in the typical case courts should treat the various members of the corporate group as the separate corporations they are and not as one client.
The community-of-interest rationale does not fit as a matter of black-letter law because the community-of-interest privilege only comes into play when parties are represented by separate counsel, which often is not the case for parents and subsidiaries. See Part IV.C.2, supra. Moreover, the community-of-interest privilege only applies when those separate attorneys disclose information to one another, not when parties communicate directly. Id. Finally, it assumes too much to think that members of a corporate family necessarily have a substantially similar legal interest (as they must for the community-of-interest privilege to apply, see id.) in all of each other’s communications. Thus, holding that parents and subsidiaries may freely share documents without implicating the disclosure rule because of a deemed community of interest stretches, we believe, the community-of-interest privilege too far.
It makes the most sense, then, to rest not applying the disclosure rule to many intra-group disclosures on the ground that the members of the corporate family are joint clients. This reflects both the separateness of each entity and the reality that they are all represented by the same in-house counsel (whether that counsel typically takes up office with the parent or with a subsidiary).
2. Keeping Control of the Privilege
In its amicus brief, the Association of Corporate Counsel (“ACC”) has as a theme a parent company’s desire to keep control of the attorney-client privilege. One particular ACC (and BCE) concern is that disclosures to parent/subsidiary cross-directors risk creating broad joint representations that subsidiaries can use to invade the parent’s privilege in subsequent adverse litigation. Not so. In thinking about these situations, courts should keep in mind the definition of disclosure. For purposes of the disclosure rule, a disclosure occurs when the parent shares an otherwise confidential attorney-parent communication with an officer, director, or agent of a subsidiary in that capacity. See Del. R. Evid. 510. Courts recognize corporate officers’ and directors’ ability to sit on multiple boards by “ ‘changing] hats.’ ” United States v. Bestfoods, 524 U.S. 51, 69, 118 S.Ct. 1876, 141 L.Ed.2d 43 (1998) (quoting Lusk v. Foxmeyer Health Corp., 129 F.3d 773, 779 (5th Cir.1997)). Thus it does not break confidence to share an attorney-parent communication with an officer of the parent in her capacity as officer of the parent, even though she is also a director or officer of a subsidiary. Because this sort of information sharing is not a disclosure for purposes of the disclosure rule, courts need not bother trying to apply a joint representation analysis to save the privilege in these situations, nor should they use such “disclosures” to find a joint representation with the relevant subsidiary.
A similar concern is that courts may find too broad of a joint representation, which in a spin-off situation ends up allowing the subsidiary to invade the parent’s privilege. When, for example, in-house counsel of the parent seek information from various subsidiaries in order to complete the necessary public filings, the scope of the joint representation is typical*373ly limited to making those filings correctly. It does not usually involve jointly representing the various corporations on the substance of everything that underlies those filings.27 Contrary to the Debtors’ argument, it is permissible (subject, of course, to conflict rules) for attorneys and clients to limit the scope of a joint representation in a sophisticated manner; nothing requires construing the scope of a joint representation more broadly than the parties to it intend.
3. When Conflicts Arise
It is inevitable that on occasion parents and subsidiaries will see their interests diverge, particularly in spin-off, sale, and insolvency situations. When this happens, it is wise for the parent to secure for the subsidiary outside representation. Maintaining a joint representation for the spinoff transaction too long risks the outcome of Polycast,28 125 F.R.D. at 49, and Medcom,29 689 F.Supp. at 844 — both cases in which parent companies were forced to turn over documents to their former subsidiaries in adverse litigation — not to mention the attorneys’ potential for running afoul of conflict rules. That the companies should have separate counsel on the matter of the spin-off transaction, however, does not mean that the parent’s in-house counsel must cease representing the subsidiary on all other matters. After all, spin-off transactions can be in the works for months (or even years), and during that time it is proper (and obviously efficient) for in-house counsel to continue to represent the subsidiary (jointly or alone) on other matters.
Once conflicts begin coming to the surface, the question of when to acquire separate counsel is often difficult. As Medcom and Polycast demonstrate, from the perspective of protecting the privilege the best answer is that once the parties’ interests become sufficiently adverse that the parent does not want future controllers of the subsidiary to be able to invade the parent’s privilege, it should end any joint representation on the matter of the relevant transaction. Polycast, 125 F.R.D. at 49; Medcom, 689 F.Supp. at 844. This standard is, of course, only relevant if the parties have already begun a joint representation on that transaction; if they have *374not, then the parent has nothing to fear as far as the privilege goes (though other concerns, such as their fiduciary duties to their subsidiaries, may argue in favor of separating counsel).
In sum, in-house counsel have available numerous means to protect a parent company’s privilege. By taking care not to begin joint representations except when necessary, to limit the scope of joint representations, and seasonably to separate counsel on matters in which subsidiaries are adverse to the parent, in-house counsel can maintain sufficient control over the parent’s privileged communications.
Y. Issues on Appeal
A. Whether the Debtors Are Entitled to Documents Generated in the Course of a BCE/Teleglobe Joint Representation
The District Court ruled that any documents generated out of a BCE/Teleglobe joint representation are subject to discovery by the Debtors. BCE argues that the Court erred because the Debtors were not a party to any such joint representation. (BCE admits that its attorneys jointly represented it and Teleglobe on some matters; it denies that they jointly represented it and the Debtors.) Indeed, the Court did not enter a factual finding that any attorney jointly represented both BCE and any of the Debtors.30 Rather, the Court ruled that the Debtors were entitled to the product of any BCE/Teleglobe joint representation because BCE had conceded this point. Alternatively, it noted the Debtors, as Teleglobe’s wholly owned subsidiaries, were entitled to the disputed documents as a matter of law, and in any event Teleglobe waived the privilege for the Debtors’ benefit in the Canadian insolvency proceedings. We address all three grounds.
1. Whether BCE’s Concession in the Bankruptcy Court Prevents it from Arguing that the Debtors are not Entitled to the Disputed Documents
a. Background
The District Court determined that BCE made a binding concession that it would produce all documents arising out of any BCE/Teleglobe joint representation and waived any argument that the Debtors are not entitled to those documents. Given the convoluted nature of the record, it is important to understand what BCE argued at each stage of the litigation.
The first reference to joint representation came from BCE itself; in a related proceeding before the Bankruptcy Court, BCE opposed a motion to compel production of various documents to the Creditors’ Committee in the course of a Rule 2004 investigation. In its opposition brief, BCE stated that “BCE attorneys consulted with attorneys, officers, or employees of Tele-globe, Inc. or its subsidiaries to discuss or provide legal advice in matters where BCE and Teleglobe Inc. (or its subsidiaries) shared a common legal interest.” App. at A01110. Furthermore, BCE declared that “such common interest privilege31 will con*375tinue to exist until the Debtors file a litigation against BCE.” App. at A01110. At the hearing, BCE stipulated that its in-house counsel, Michel Lalande, held documents relating to “two types of matters.” App. at A0172. The first category of documents “was where the common interest was involved.” App at A0172. The second category comprised documents in which “Mr. Lalande was counseling ... BCE alone.” App. at A0172. The Bankruptcy Court ordered BCE to produce to the Debtors and Creditors’ Committee “the common interest documents.” App. at A0172. Indeed, BCE agreed to do so. In the course of the hearing, however, the parties never honed in on what the “common interest” was.
The issue came up again when the Debtors filed suit. In front of the District Court (before it referred discovery matters to the Special Master), BCE stated in its initial brief in opposition to the motion to compel production that “documents were produced pursuant to the Bankruptcy Court’s Order because they were subject to the common interest privilege. Those documents were shared with [various attorneys and accountants representing the Debtors].” App. at A0218. At the first hearing before the District Court, BCE described the situation thus: “[T]here were some places where BCE attorneys clearly worked with what they conceived of as the common interest of Teleglobe and BCE. And there were areas of common interest. Judge Walrath said ... produce them.... We did.” App. at A00264. Again, none of these statements sets out a definition or clarified the perceived scope of the “common interest.”
Another round of briefing followed the District Court’s reference to the Special Master. At a hearing before the Special Master, the Debtors’ attorney went through BCE’s Rule 2004 privilege log and summarized that BCE claimed a “common-interest privilege” on documents covering the following subjects: “Tele-globe financing, financial transaction between Teleglobe and BCE, issuance and redemption of preferred shares ... [,][r]e-financing of bank facilities ..., developing Teleglobe financial statements, ... public disclosures, .... [and] Teleglobe restructuring.” App. at A00496-97. He further stated that “[BCE] claims a common interest on Project X, and so on.... [H]aving claimed common interests and having assigned lawyers to effectively represent both sides of the common interest, it would seem to us to be inconsistent now to say that we can’t have all documents relating to the common interests[,] particularly when we’re talking about documents that were reviewed and considered by our fiduciaries.” App. at A00497.
BCE responded that its in-house counsel provided limited legal services to Tele-globe on an ad hoc basis and that BCE’s in-house counsel did not serve the Debtors at all because they had their own counsel in Reston, Virginia. App. at A00497. As to Project X, BCE responded that in-house counsel Michel Lalande and Martine Tur-cotte provided very limited advice to Tele-globe on its public filings, its management representation letters, its board resolutions, and its engagement letter with an investment bank. App. at A00498. Moreover, BCE argued that this advice was limited in scope and all other Project X advice was provided solely to and for the benefit of BCE. In sum, BCE argued that “what [is reflected on the privilege log] is BCE-privileged advice, not advice ... where BCE has a common interest with [Teleglobe], not advice given to [Teleglobe] or [the Debtors,] if there is any.” App. at A00503.
*376Recognizing that there was a dispute whether the documents on the privilege log reflected advice provided solely to BCE (and thus outside of the scope of any joint representation), the Special Master ordered an audit of 50 documents to “see if I can satisfy myself that they are what you [BCE] term exclusively related to advice provided to BCE.” App. at A00504. To clarify the issue, the Special Master asked: “[YJou’ve [BCE] given them [the Debtors] the documents that deal with advice provided [to Teleglobe], by lawyers who might have been representing both BCE and [Teleglobe] at the same time, but have not given them documents that may have addressed [Teleglobe], but were only involved with advice given to BCE?” App. at A00505. BCE counsel responded, “That’s correct.” App. at A00505.
After the Special Master ordered production of all of the documents, BCE clearly raised, in its brief in opposition to the Debtors’ motion in the District Court to affirm and adopt the Special Master’s decision, the issue of its obligation to turn over documents arising from its and Teleglobe’s joint representation by BCE in-house counsel. App. at A00826. This was probably a reaction to the Special Master’s finding that the joint representation was so broad in scope that it covered all advice rendered on Project X. In other words, BCE had not objected to producing the “common interest” documents so long as its understanding of the narrow scope of the common interest prevailed, but when the Special Master found a broad scope, it asserted that it need not produce the common-interest documents at all. The District Court rejected this argument out-of-hand, finding that BCE was bound by its concession in the Bankruptcy Court that it would produce to the Debtors and the Creditors’ Committee all of the documents of common interest to BCE and Tele-globe.32
b. Merits
BCE should not, the Debtors argue, be able to raise now the issue of whether they and BCE were part of a joint representation. They base this contention on issue waiver, judicial admission, and judicial es-toppel grounds. We address each, as well as whether BCE’s waiver of the privilege in the Bankruptcy Court prospectively waived it as to any documents found to be within the scope of a BCE/Teleglobe joint representation.
i. Issue Waiver
Our longstanding rule is that a party must raise an issue before the District Court in order to press it on appeal.33 To raise an issue, a party must present it with sufficient specificity to allow the court to pass on it. See Keenan v. City of Philadelphia, 983 F.2d 459, 471 (3d Cir. 1992); accord Shell Petroleum, Inc. v. United States, 182 F.3d 212, 218 (3d Cir. 1999) (“A party ... must unequivocally put its position before the trial court at a *377point and in a manner that permits the court to consider its merits.”)- At oral argument before the Special Master, BCE stated unequivocally that its in-house attorneys never represented the Debtors on anything related to Teleglobe’s restructuring. The Special Master was skeptical of this assertion, and asked the Debtors to respond to it. They stated that they were entitled to documents generated through a BCE/Teleglobe joint representation because they, as third-level wholly owned subsidiaries, were members of the same community of interest (once again, an instance where someone conflates the community-of-interest privilege with the co-client privilege). This exchange properly put before the Special Master the issue of whether the Debtors were (or needed to be) parties to the BCE/Teleglobe joint representation for production to be ordered. Moreover, BCE again raised the issue in its objections to the Special Master’s decision. App. at A00826. With this context, we cannot conclude BCE waived contending that it and the Debtors were not jointly represented on any matter relevant to this case.
ii. Judicial Admission
The Debtors argue, however, that BCE’s statement to the Bankruptcy Court that it would turn over to the Debtors (and the Creditors’ Committee) the “common interest” documents arising out of the BCE/Teleglobe joint representation was a binding judicial admission. To be binding, admissions must be unequivocal. Glick v. White Motor Co., 458 F.2d 1287, 1291 (3d Cir.1972). Similarly, they must be statements of fact that require evidentiary proof, not statements of legal theories. Id.
Applying those principles, BCE’s statements on any joint representation of BCE and the Debtors were never unequivocal. All references to a joint representation of BCE and the subsidiaries were couched in alternative language. See, e.g., App. at A01110 (“BCE attorneys consulted with attorneys, officers, or employees of Tele-globe, Inc. or its subsidiaries to discuss or provide legal advice in matters where BCE and Teleglobe, Inc. (or its subsidiaries) shared a common legal interest.” (emphases added)). To the extent that BCE admitted that it was obligated to turn over documents related to a joint representation on a matter of common interest, such an admission appears more like a statement of a legal theory or position than a statement about an issue of fact. Thus the judicial admission doctrine is not applicable.
iii. Judicial Estoppel
Judicial estoppel prevents a party from “playing fast and loose with the courts” by adopting conflicting positions in different legal proceedings (or different stages of the same proceeding). Delgrosso v. Spang & Co., 903 F.2d 234, 241 (3d Cir.1990) (internal quotation marks and citations omitted). Here, the Debtors claim that BCE’s position that its attorneys never jointly represented it and the Debtors is impermissible because BCE relied on the joint representation in the Bankruptcy Court to assert a claim of privilege and because it induced the Bankruptcy Court and Special Master to rely on its concession that common-interest documents had been produced, only changing its position when the Special Master found its representations untrue. Judicial estoppel requires (1) a clear inconsistency and (2) that the party estopped obtain an unfair advantage from that inconsistency. In re Armstrong World Indus., Inc., 432 F.3d 507, 517-18 (3d Cir.2005). This case looks more like a legitimate disagreement over the scope of any joint representation (mixed with a dose of sloppiness) than it does a bad faith attempt to mislead the courts. BCE’s position throughout has *378been that it has turned over the documents that arose out of a BCE/Teleglobe joint representation. It told the Bankruptcy Court — in equivocal terms — that there may have been a BCE/Debtors joint representation (and if so, it agreed to turn over any such documents). In front of the Special Master, BCE’s counsel specifically stated that its attorneys did not jointly represent the Debtors, and it has maintained that position since. In this limited context, that BCE was “playing fast and loose” with the courts appears too strong a statement. Perhaps it was being hyper-technical, and indeed up against the “too close for cricket” line, but that is as far as we can conclude on the record before us.
iv. Implied Prospective Waiver of the Privilege
To repeat, in the Bankruptcy Court BCE agreed to produce to the Debtors and the Creditors’ Committee what it termed the “common interest” documents to resolve a discovery dispute. It did not explain the scope of the common interest, but it did agree to produce the documents. BCE views the Debtors’ contention that it may not now assert the privilege over other documents alleged to arise from the same joint representation as an implied prospective waiver argument. That analysis makes some sense.
In discovery disputes, implied waivers are construed narrowly, In re Lott, 424 F.3d 446, 453 (6th Cir.2005), and a party is only forced to produce documents under a prospective waiver theory if it agrees to disclose only favorable privileged documents while keeping for itself the unfavorable ones to gain an advantage in litigation. See Westinghouse, 951 F.2d at 1426 n. 12 (“When a party discloses a portion of otherwise privileged materials while withholding the rest, the privilege is waived only as to those communications actually disclosed, unless a partial waiver would be unfair to the party’s adversary. If partial waiver does disadvantage the disclosing party’s adversary by, for example, allowing the disclosing party to present a one-sided story to the court, the privilege will be waived as to all communications on the same subject.”); accord Tackett, 653 A.2d at 260 (holding that Delaware courts will only find a prospective waiver when a party uses partial disclosure as a weapon).
On the record here, it appears that BCE agreed to produce documents that fell within its understanding of the BCE/Tele-globe joint representation, not the masses of documents that the Special Master eventually found to fall within that category. Moreover, the Debtors have not argued that BCE is seeking an improper benefit by selectively disclosing documents. Thus we find no implied prospective waiver in favor of the Debtors as to whatever documents a court might conclude were part of a BCE/Teleglobe joint representation.
2. Whether the Community-of-Interest Privilege Entitles the Debtors to the Documents as a Matter of Law
The District Court also ruled that the Debtors were entitled to the documents as a member of BCE’s community of interest. But as explained in Part IV.F. 1, supra, it is not the case that parents and subsidiaries are in a community of interest as a matter of law. Moreover, the community-of-interest privilege only applies to parties represented by separate counsel, see Part IV.C.2, supra. Even some of the cases that the Debtors cite suggest more nuance than they admit. See, e.g., In re S. Air Transp., Inc., 255 B.R. 706, 711 (Bankr.S.D.Ohio 2000) (“Parent and subsidiary corporations generally share a common interest, and may, in appropriate circumstances, be considered a single *379client for purposes of the attorney-client privilege.” (emphases added)). The majority — and more sensible — view is that even in the parent-subsidiary context a joint representation only arises when common attorneys are affirmatively doing legal work for both entities on a matter of common interest. See, e.g., Polycast, 125 F.R.D. at 49 (finding a joint representation when a parent’s officer and general counsel affirmatively advised subsidiary on how to comply with merger agreement to which parent and subsidiary were both parties). A broader rule would wreak havoc because it would essentially mean that in adverse litigation a former subsidiary could access all of its former parent’s privileged communications because the subsidiary was, as a matter of law, within the parent entity’s community of interest.
3. Whether Teleglobe’s Waiver of the Privilege for the Debtors’ Benefit in the Canadian Insolvency Proceedings Entitles them to the Documents
The Debtors argue that they are entitled to any documents generated in the course of the BCE/Teleglobe joint representation because Teleglobe, through its Plan Administrator in the Canadian insolvency proceedings, has waived the attorney-client privilege in their favor.34 The question before us is whether one party to a joint representation (Teleglobe) may unilaterally waive the privilege and thereby force the other (BCE) to turn over documents generated in the course of the joint representation to third parties (the Debtors). Put differently, the question is whether the co-client privilege is subject to a unilateral control rule (either co-client may waive the privilege unilaterally and thus force the other to turn over documents produced during the course of the joint representation to third parties) or a bilateral control rule (both clients must agree to waive the privilege in order for the waiver to take effect). The general answer is bilateral control. As explained in Part IV.D, supra, this rule is in the Restatement:
[I]n the absence of an agreement with co-clients to the contrary, each co-client may waive the privilege with respect to that co-client’s own communications with the lawyer, so long as the communication relates only to the communicating and waiving client. One co-client does not have authority to waive the privilege with respect to another co-client’s communications to their common lawyer. If a document or other recording embodies communications from two or more co-clients, all those co-clients must join in a waiver, unless a nonwaiving co-client’s communication can be redacted from the document.
Restatement (Third) of the Law Governing Lawyers § 75 cmt. e. The Delaware District Court has predicted that Delaware law accords with this principle. Interfaith Hous. of Del., 841 F.Supp. at 1402, and the Delaware Court of Chancery (albeit in an unpublished decision) has agreed. Tenneco Auto., Inc. v. El Paso Corp., 2001 WL 1456487, at *2 (Del.Ch.2001). Canadian law on this point is the same. Ashburton Oil Ltd. v. Sharp, 67 B.C.L.R.2d 64, ¶24 (B.C.1992).
The cases the Debtors cite are not explicitly contradictory. In re Grand Jury Subpoenas, 902 F.2d 244, 248 (4th Cir. 1990), for instance, deals with the transfer *380of the subsidiary’s own privilege from old to new management, not with joint representation. Three other cases the Debtors cite — Bass Public Co. v. Promus Cos., 868 F.Supp. 615 (S.D.N.Y.1994), Polycast, and Medcom — are muddier, as there are statements in all three that seem to disagree with the Restatement. See Bass, 868 F.Supp. at 621 (stating that the bilateral control rule only applies if the two clients are actual or potential co-defendants); Polycast, 125 F.R.D. at 50 (same); Medcom, 689 F.Supp. at 844-45 (same).35 At the same time, none of the three decisions actually violated the Restatement. In Bass, one of the plaintiffs was the subsidiary (ie., the joint client); it therefore was entitled to everything generated in the course of the joint representation as a matter of right. Bass, 868 F.Supp. at 617. In Polycast and Medcom, the disputed communications were between the subsidiary and the joint attorneys; thus the subsidiary could waive the privilege by itself. Medcom, 689 F.Supp. at 844; Polycast, 125 F.R.D. at 48. Here, however, we have third parties seeking documents produced for BCE as (allegedly) part of a BCE/Teleglobe joint representation. Under these circumstances, we predict that the Delaware Supreme Court would follow the Restatement rule (as accepted by both the Delaware Court of Chancery and the federal District of Delaware Court) that Tele-globe alone cannot waive the privilege as to communications involving BCE.
4. Conclusion and Remand
Ordering the production of documents on the privilege log must be predicated on a factual finding that BCE and the Debtors were parties to a joint representation. Because there is no such finding on record, we remand for further factfinding on this issue. On remand, that factfinding should consider § 14 of the Restatement, which details how a lawyer/client relationship arises, and our discussion in Part IV.C, supra, on the existence and scope of a co-client relationship.
B. The Effect of Funneling Documents Through BCE’s In-House Counsel
The District Court held that because BCE funneled all of the contested documents through in-house counsel who were jointly representing Teleglobe, all of those documents are discoverable to the Debtors as part of the BCE/Teleglobe joint representation. This ruling applied even to documents produced by outside counsel that did not represent Teleglobe (or the Debtors). As an initial matter, the District Court’s reasoning cannot survive our determination in section A of this Part that whether BCE and Teleglobe were jointly represented does not matter without an *381additional finding that the Debtors were also parties to that representation. Still, we examine the Special Master and District Court’s reasoning because it may be relevant on remand.
In reaching his decision on this issue, the Special Master relied primarily on In re Mirant Corp., 326 B.R. 646, 651 (Bankr. N.D.Tex.2005). In that case, a parent company and its wholly owned subsidiary formally engaged the Troutman Sanders law firm to represent them jointly with regard to the sale of 20% of the subsidiary’s stock. Id. at 648. The Court held that the subsidiary/debtor could obtain documents arising out of the joint representation despite the fact that the engagement letter stipulated that the parties agreed that their individual communications with the joint attorneys would be privileged against one another. Id. at 652. That holding is off point here because it did not speak to whether documents outside of the scope of the joint representation and provided by outside attorneys are brought within that scope merely because they pass through in-house counsel who are jointly representing the subsidiary as well as the parent; rather, it dealt only with documents produced within the scope of the joint representation by the joint attorneys.
Here, the Eureka principle seems most apt: when BCE took the trouble of hiring outside counsel to provide advice only to it on divestiture issues, it reasonably expected that those communications would be privileged. Whether BCE in-house counsel should have refused to look at the documents because of their own conflicts is, under Eureka, beside the point, as is whether its in-house counsel had a fiduciary duty to share information with Teleglobe. 743 F.2d at 937-38. (Not to be forgotten is that BCE’s in-house counsel also had a duty to keep its communications confidential.)
The guiding principle of Eureka is that when an attorney errs by continuing to represent two clients despite their conflicts, the clients—who reasonably expect their communications to be secret— are not penalized by losing their privilege. Id. Indeed, Eureka is merely one in a line of cases that hold that communications outside the scope of the joint representation or common interest remain privileged. See, e.g., CSX Transp., Inc. v. Lexington Ins. Co., 187 F.R.D. 555, 560 (N.D.Ill.1999) (holding that documents outside the scope of the common interest need not be disclosed); Stratagem, 153 F.R.D. at 543 (same); Pittston Co. v. Allianz Ins. Co., 143 F.R.D. 66, 69-71 (D.N.J.1992) (same); Carey-Canada Inc. v. Aetna Cas. & Sur. Co., 118 F.R.D. 250, 251 (D.D.C.1987) (same).
The District Court relied primarily on policy analysis: it noted that forcing BCE to choose between (1) allowing in-house counsel to filter the information from outside counsel while losing the privilege, and (2) maintaining the privilege by not working directly with outside counsel, is “harsh.” App. at A0015. It stated, however, that the result was required by the law of the adverse-litigation exception to the joint-client privilege, and that BCE could have protected itself by clearly terminating any representation of Teleglobe by its in-house lawyers when it began reconsidering its funding of Teleglobe.
The Mirant Court also engaged in policy analysis that the Debtors endorse:
Even were there merit in the arguments of Troutman and [the parent company] TSC, the court would be most reluctant to deny Debtors the requested discovery for reasons of public policy. It is black-letter law that the attorney-client privilege is meant to foster open communications between attorney and client.... In a bankruptcy case, the *382need for investigation is far more acute than is any concern for attorney-client communications.... Debtors, acting as fiduciaries for the benefit of their creditors, are pursuing an investigation which is important not only to those who may have lost money as a result of Debtors’ demise. It is critical that both those who purchased Mirant’s (and its subsidiaries’) securities and the public have confidence that potential liability of TSC (and Troutman) has been thoroughly explored. That Debtors sought chapter 11 relief less than two and one half years after TSC completed their divestiture is reason enough to raise concern that all might not have been right in the transactions between TSC and Mirant.... Given the short time between divestiture and commencement of these chapter 11 cases and given the pre-divestiture history of Debtors’ problems, it is essential to the integrity of the chapter 11 process that no stone be left unturned in ensuring satisfactory completion of Debtors’ investigation.
326 B.R. at 654-55 (citations omitted).
In a similar vein, the Debtors argue that BCE created and controlled this situation, ie., it tried to have it both ways by having in-house counsel work primarily for BCE while still jointly representing Teleglobe. Because of its control over the situation, the Debtors argue, BCE should not be heard to complain about the legal ramification of losing its privilege against producing documents. They further contend that BCE could have obtained outside counsel for Teleglobe at any time and ceased its joint representation, and because it did not do so, it must relinquish its documents.
As noted, the District Court here conceded that this result was harsh, as did the Mirant Court. Both, however, seemed to believe that allowing the privilege to stand would make it too easy for parent companies to hide their wrongdoing. This truth-seeking rationale, however, is a problem because the privilege is admittedly not truth-seeking. United States v. Liebman, 742 F.2d 807, 810 (3d Cir.1984) (“It does not advance resolution of the issue to argue ... that the attorney-client privilege is an obstacle to the search for the truth.” (internal citations and quotations marks omitted)). Moreover, Delaware abrogates the privilege in the face of serious wrongdoing facilitated by the attorney through the crime/fraud exception. Del. R. Evid. § 502(d)(1) (“There is no privilege under this rule ... [i]f the services of the lawyer were sought or obtained to enable or aid anyone to commit or plan to commit what the client knew or reasonably should have known to be a crime or fraud.”). This reflects a policy choice that, absent well-founded allegations not only of serious wrongdoing but of serious wrongdoing facilitated by the attorney, the privilege should stand. Here, the Debtors do not appear to have alleged that BCE’s attorneys aided any wrongdoing. Moreover, invoking the crime-fraud exception requires that the plaintiff make a prima facie showing of evidence supporting the elements of the exception. In re Grand Jury Investigation, 445 F.3d 266, 274 (3d Cir.2006). While that showing is not onerous, it is necessary, for “a mere charge of wrongdoing will [not] ... put the privilege to flight.” Clark v. United States, 289 U.S. 1, 14, 53 S.Ct. 465, 77 L.Ed. 993 (1933) (Cardozo, J.). The adverse-litigation exception to the joint-client privilege should not be a back-door means of destroying the attorney-client privilege on the basis of alleged wrongdoing; if the Debtors had wanted to lay to rest the privilege on those grounds, they should have asserted the crime-fraud exception directly and allowed the District Court to pass on whether they could make the required prima facie showing.
In addition, Eureka counsels that it is the scope of the joint representation — not *383whether communications were shared with joint attorneys — that is dispositive. 743 F.2d at 937-38. If the communications were outside the scope of the joint representation, then sharing them with a conflicted joint attorney is of no moment, even if the conflicted attorney acted improperly in accepting the communications. Id. If the communications were within the scope, then they are discoverable. Id. at 937. Either way, the fact that documents prepared by outside counsel were shared with in-house counsel does not have the significance that the District Court and Special Master attached to it. What we believe is significant is that, on remand, the District Court needs to determine whether any attorneys jointly represented BCE and the Debtors on a matter of common interest. If they did, then any documents within the scope of that joint representation are discoverable.
Amicus curiae ACC fears that invoking Eureka implies that in-house counsel act improperly when they review documents from outside counsel. Not so. Here we have a finding of fact that BCE’s in-house counsel jointly represented BCE and Tele-globe on all issues relating to Teleglobe’s ultimate spin-off. If that finding is correct, then it was bad judgment for BCE’s in-house attorneys to take part in outside counsel’s separate representation of BCE on the same issues if it still wanted the shield of the attorney-client privilege. We note, however, that the factual finding gives us pause. As we detailed in Part IV.F, supra, parent-subsidiary joint representations are typically quite narrow in scope. Still, we have not reviewed all of the documents that the Special Master and District Court did, so we can neither affirm nor vacate that finding on the record before us — and we need not because the Debtors’ status as third parties to any BCE/Teleglobe joint representation makes its scope irrelevant as a matter of law.
We agree with ACC that a rule forcing parent companies to choose between relinquishing the privilege on one hand and relinquishing any control over their subsidiaries makes little sense. If the parent chooses to forgo the use of its in-house counsel on an important transaction, then it loses the advisors that know the most about its legal health. If it chooses to cut off its subsidiary, then it risks liability when the subsidiary and parent do not operate in tandem. Putting that choice in the context of this case, it appears that BCE spent some four months deciding what to do with Teleglobe. Over the course of Project X, it considered a variety of options that did not involve abandoning Teleglobe. Moreover, during this time it was still responsible for reviewing Tele-globe’s public filings. Thus for BCE’s in-house counsel to cut Teleglobe off would have been an expensive and risky proposition. Similarly, given that BCE was in the process of making a very serious business decision with important legal implications, to deprive it of its in-house counsel simply so that those attorneys could continue to advise Teleglobe on other matters seems overly harsh.
To prevent this outcome, it is important for in-house counsel in the first instance to be clear about the scope of parent-subsidiary joint representations. By properly defining the scope, they can leave themselves free to counsel the parent alone on the substance and ramifications of important transactions without risking giving up the privilege in subsequent adverse litigation.
VI. Potential Alternate Sustaining Grounds
A. The Fiduciary Exception to the Attorney-Client Privilege
Before the Special Master and in supplemental briefing, the Debtors argue that *384they should be allowed access to the disputed documents under the fiduciary exception to the attorney-client privilege as articulated in Garner v. Wolfinbarger, 430 F.2d 1093 (5th Cir.1970). The core holding of Gamer is as follows:
The attorney-client privilege still has viability for the corporate client. The corporation is not barred from asserting it merely because those demanding information enjoy the status of stockholders. But where the corporation is in suit against its stockholders on charges of acting inimically to stockholder interests, protection of those interests as well as those of the corporation and of the public require that the availability of the privilege be subject to the right of the stockholders to show cause why it should not be invoked in the particular instance.
Garner, 430 F.2d at 1103-04. Garner thus allows shareholders of a corporation to invade the corporation’s privilege in order to prove fiduciary breaches by those in control of the corporation upon showing good cause.
As the Debtors note, while the Delaware Supreme Court has never adopted Gamer, the Court of Chancery has adopted and applied it in numerous decisions. Indeed, Vice Chancellor Lamb wrote that “Delaware courts follow the approach outlined in Garner.” Grimes v. DSC Comm. Corp., 724 A.2d 561, 568 (Del.Ch.1998) (citing Deutsch v. Cogan, 580 A.2d 100, 105 (Del. Ch.1990) (“Delaware courts have consistently followed Gamer.”)).
In Deutsch, the Court of Chancery extended Gamer to a situation in which cashed-out minority shareholders of a subsidiary sought to abrogate the attorney-client privilege of the parent company’s controlling shareholder in a dispute over a merger transaction that the minority shareholders contended was not fundamentally fair. 580 A.2d at 102-03. The Court held that Garner applied by analogy because the parent’s controlling shareholder, as the entity in fact controlling the subsidiary, owed the minority shareholders fiduciary duties. Id. at 107. Therefore, in a dispute over compliance with those duties, the minority shareholders could— upon showing good cause — overcome the privilege with regard to documents related to the challenged transaction. Id. at 108.
The Debtors’ argument here is similar: because BCE controlled the Debtors while they were insolvent, it owed fiduciary duties to the Debtors of which their creditors (not BCE) were the primary beneficiaries. See generally N. Am. Catholic Programming Found., 930 A.2d at 101-02, 2007 WL 1453705, at *7. Thus, in this dispute, in which the Debtors are asserting a breach of those duties for the benefit of their creditors, the Debtors should be able to put aside the privilege upon showing good cause. It is possible that Delaware courts would extend the Gamer fiduciary exception to a situation like this one. The key, however, is insolvency. If the Debtors were not insolvent (or in that hazy “zone of insolvency,” see supra note 9) at the time of the otherwise privileged communications, then BCE was the only beneficiary of the Debtors’ success, and it could not, therefore, breach its fiduciary duty to itself.
Whether a corporation is insolvent, and when it becomes so, are issues of fact. Trenwick, 906 A.2d at 195. Under Delaware law, a corporation is insolvent if it has: “1) a deficiency of assets below liabilities with no reasonable prospect that the business can be successfully continued in the face thereof, or 2) an inability to meet maturing obligations as they fall due in the ordinary course of business.” Prod. Res. Group, L.L.C. v. NCT Group, Inc., 863 A.2d 772, 782 (Del.Ch.2004) (Strine, V.C.) (internal citations and quotation marks omitted). Here, both parties seem *385to agree that the Debtors were all deeply in debt with no significant revenue stream other than BCE’s funding. The dates on which each Debtor became insolvent, however, are issues of fact that we cannot resolve here.
In addition, the Debtors' argument appears to have a significant deficiency: they must have a colorable claim of breach of fiduciary duty to show “good cause.” Garner, 430 F.2d at 1104. Even if BCE owed the Debtors and their creditors fiduciary duties, it seems a stretch to argue that BCE’s decision not to fund Teleglobe implicated those duties. A fiduciary ordinarily has the obligation (protected by the business judgment rule) to manage the affairs of a corporation in such a way as to maximize its economic value; it does not have a duty to guarantee or bail out a corporate family member when it loses money. Trenwick, 906 A.2d at 205. As Vice Chancellor Strine noted in Trenwick, insolvency does not render a corporation’s fiduciaries guarantors of that corporation’s success; rather, it merely expands the universe of people with standing to assert a beneficial interest in the fiduciaries’ obligation to maximize the value of the corporation.36 Id. Whether BCE would continue to infuse Teleglobe and the Debtors with funding seems, at least at first glance, entirely BCE’s decision. Still, we do not have the entire record before us, and so we leave this issue for the District Court to resolve in the first instance.
Mopping up, BCE advances seven arguments against applying Gamer in this situation, most of which are unavailing. First, it argues that Gamer does not apply to work product. This is correct, but as we do not know how many of the 800 different documents currently in dispute contain work product, we must leave this issue for the District Court to resolve on remand. Second, BCE (the appellant) argues that the Debtors (the appellees) have abandoned the Gamer argument. The cases it cites, however, are off the mark because they address the appellant’s obligation to raise all grounds for reversal. See Simmons v. City of Philadelphia, 947 F.2d 1042, 1066 (3d Cir.1991) (Opinion of Becker, J., announcing the judgment of the Court); Inst. for Scientific Info., Inc. v. Gordon & Breach, Sci. Publishers, Inc., 931 F.2d 1002, 1011 (3d Cir.1991). It is firmly established that we may affirm, on any ground supported by the record, so the Debtors’ failure to raise the issue does not waive it. Azubuko v. Royal, 443 F.3d 302, 303 (3d Cir.2006). Moreover, the issue has yet to be addressed by the District Court, and it remains open on remand.
Third and fourth, BCE argues that Gamer has been rejected by Canadian courts and never adopted by our Court. As a matter of federal common law, it is correct that we have not always applied Garner, see Wachtel v. Health Net, Inc., 482 F.3d 225, 238 (3d Cir.2007), but neither party argues that federal common law governs this dispute.37 As to the Canadian *386law argument, we doubt that Canadian law would apply to this issue, as the Debtors are Delaware corporations.38 Under the internal affairs doctrine, anyone controlling a Delaware corporation is subject to Delaware law on fiduciary obligations to the corporation and other relevant stakeholders. See In re Topps Co. S’holders Litig., 924 A.2d 951, 960 (Del.Ch.2007) (Strine, V.C.) (explaining that the law of fiduciary obligations is one of the most important ways a state regulates a corporation’s internal affairs); Restatement (Seoond) of Conflict of Laws § 306. Moreover, if Delaware law favors admission of the evidence, then BCE must show some “special reason” why that should not be given effect. Id. § 139(2); see also Carlton Invs., Inc. v. TLC Beatrice Int'l Holdings, Inc., No. C.A. 13950, 1996 WL 33167792, at *2 (Del.Ch. Sept.27, 1996). Still, because we ultimately conclude that we cannot apply Gamer here without knowing when the Debtors became insolvent, we will not conduct a full-scale choice-of-law analysis.
Fifth and sixth, BCE argues that Garner cannot be extended to this situation because (1) BCE owed the Debtors no fiduciary duties, and (2) Gamer does not permit subsidiaries to invade the privilege of indirect corporate parents. The answer to both arguments is that BCE, as the ultimate owner of more than half (and, indeed, all) of the Debtors’ voting power, owed them the duties of care and loyalty. See Weinstein Enters. v. Orloff, 870 A.2d 499, 507 (Del.2005). If the Debtors were solvent, then all duties flowed back up to BCE as the only party with a legitimate interest in the Debtors’ success. If that were the case, then BCE is correct that it effectively owed the Debtors no duties. However, if the Debtors were insolvent, then their creditors also had a legitimate interest in their success. See N. Am. Catholic Programming Found., 930 A.2d at 101-02, 2007 WL 1453705, at *7. With multiple stakeholders, BCE’s duties of care and loyalty would come into play in the same way that the directors’ duties did in Garner, and its attorney-client privilege could be set aside by showing good cause.
Finally, BCE argues that the Debtors cannot show good cause. As BCE concedes in the next breath, however, it is not for us to determine whether they can show good cause in the first instance; so we leave the issue open on remand.
B. Affirming as a Discovery Sanction
The Debtors urge us to affirm the District Court’s order as a sanction for BCE’s penchant for designating too many documents as privileged. We cannot do so at this time for the simple reason that the District Court did not impose such a sanction. To be sure, both the Special Master and District Court expressed displeasure with BCE’s litigation conduct, but neither expressly penalized that behavior, and so we have no sanction order to review. We do note that preventing a party from asserting the attorney-client privilege is a legitimate sanction for abusing the discovery process, and we do not foreclose that remedy on remand. Disclosure is a serious sanction, but one that may be imposed only if the District Court finds bad faith, wilfulness, or fault. See Am. Nat'l Bank & Trust Co. of Chicago v. Equitable Life Assur. Soc’y of U.S., 406 F.3d 867, 877-80 (7th Cir.2005).
VII. Conclusion
We hold that the District Court may only compel BCE to produce disputed *387documents because of the adverse-litigation exception to the co-client privilege, see Part IV.D, supra, if it finds that BCE and the Debtors were jointly represented by the same attorneys on a matter of common interest that is the subject-matter of those documents. Finding that BCE and Tele-globe were jointly represented is not enough, as Teleglobe cannot unilaterally waive the co-client privilege that attaches to documents that involve BCE and were created in the course of the joint representation. Moreover, BCE has not waived the argument, and it is not in some “community of interest” with the Debtors as a matter of law. In addition, that documents prepared by outside counsel were tunneled through in-house counsel for both BCE and Teleglobe is of no moment. Following Eureka, what matters is the scope of any joint representation: documents within the scope are discoverable; documents outside it are not, irrespective of whether they were improperly tunneled through joint attorneys.
On remand, then, the primary issue is whether any attorneys jointly represented BCE and the Debtors on a matter of common interest. Also open on remand are the issues of discovery sanctions and whether the Garner fiduciary exception— currently extant in Delaware — applies in this case.
Because of the need to resolve this privilege dispute efficiently so that the underlying litigation can proceed, this panel of our Court will continue to review any additional privilege-related appeals.