Paul Biddle appeals the district court’s dismissal of his cause of action for lack of subject matter jurisdiction. Biddle brought a qui tam lawsuit against the Board of Trustees of the Leland Stanford, Jr. University (“Stanford”) pursuant to the False Claims Act (“FCA”), 31 U.S.C. § 3730 , alleging that Stanford defrauded the United States Government. Because the district court lacks subject matter jurisdiction over Biddle’s lawsuit, we affirm.

BACKGROUND

Federal government agencies regularly enter into agreements with Stanford and other universities for research and other activities to be performed by the universities. The agreements are designed to reimburse the universities for both direct and indirect costs, as well as staff benefits. Direct costs include salaries and supplies for a particular project. Indirect costs are a method' to allow a university to be reimbursed for a share of its overhead, such as costs of buildings, equipment, utilities, and administrative support. The Office of Naval Research (“ONR”) is responsible for setting indirect cost rates and staff benefits at Stanford. The agreements between ONR and Stanford for calculating-indirect costs are called Memoranda of Understanding.

In October of 1988, Biddle was hired as Administrative Contracting Officer (“ACO”) and Resident Representative for ONR at Stanford. Soon thereafter, Biddle came to believe that Stanford was overcharging the government for indirect costs. After informing his supervisors to no avail, Biddle relayed his concerns to a congressional subcommittee in the summer of 1990. As a result of Biddle’s revelations, the General Accounting Office and the Defense Contract Audit Agency began an investigation of Stanford.

In September of 1990, the media began reporting on Biddle’s allegations against Stanford. Biddle was interviewed by numerous newspapers and magazines, and was featured on the ABC news program “20/20.” In 1991, Stanford’s indirect cost rate was reduced from 76% to 55.5%.

On September 9,1991, after more than one year of extensive media coverage of Stanford’s alleged overcharges, Biddle filed a qui tam suit under the False Claims Act. Follow- mg a two-year investigation, the Department of Justice decided not to intervene in Biddle’s lawsuit. Biddle’s complaint was then served on Stanford. On August 30, 1995, Biddle filed his third amended complaint. Stanford moved to dismiss the complaint for lack of subject matter jurisdiction. On August 26, 1996, the district court granted Stanford’s motion. Specifically, the district court held that Biddle’s complaint was jurisdietionally barred because (1) it was based upon public disclosures, and (2) Biddle did not qualify as an original source of the information he provided to the government.

STANDARD OF REVIEW

The existence of subject matter jurisdiction is a question of law reviewed de novo. Ma v. Reno, 114 F.3d 128, 130 (9th Cir.1997). This court also reviews de novo the district court’s conclusion that it lacks subject matter jurisdiction. H2O Houseboat Vacations, Inc. v. Hernandez, 103 F.3d 914, 916 (9th Cir.1996). In making its determination, the district court may resolve factual disputes based on the evidence presented where the jurisdictional issue is separable from the merits of the case. Rosales v. United States, 824 F.2d 799, 803 (9th Cir.1987). The district court’s findings of fact relevant to its determination of subject matter jurisdiction are reviewed for clear error. Id.

DISCUSSION

Under the False Claims Act, any person who defrauds the United States Government is liable for civil penalties. 31 U.S.C. § 3729 (1994). Although the FCA requires the Attorney General to investigate possible violations, id. § 3730(a), the FCA also permits civil qui tam actions by private persons, known as relators, id. § 3730(b). In a qui tam action, the relator sues on behalf of the government as well as himself. If the relator prevails, he receives a percentage of the recovery, with the remainder being paid to the government. Congress, however, has limited the jurisdiction of courts over qui tam actions:

Id. § 3730(e)(4)(A). “Original source” is defined as “an individual who has direct and independent knowledge of the information on which the allegations are based and has voluntarily provided the information to the Government before filing an action under this section which is based on the information.” Id. § 3730(e)(4)(B).

I. Were the disclosures of Stanford’s alleged fraud “public disclosures” under the FCA?

Biddle first argues that the district court erred in ruling that the disclosures of Stanford’s alleged fraud through governmental investigations and media reports constituted “public disclosures” under the FCA. Biddle relies on United States ex rel. Barajas v. Northrop Corp., 5 F.3d 407 (9th Cir.1993), ce rt. denied, 511 U.S. 1033 , 114 S.Ct. 1543 , 128 L.Ed.2d 195 (1994), for the proposition that if an individual provides information to the government that causes a governmental investigation, and evidence of fraud or wrongdoing is made public during the government’s investigation, then allegations regarding the fraud are not treated as publicly disclosed under the FCA.

In Barajas, the plaintiff brought a qui tam suit against Northrop for fraudulent activities. Following Northrop’s indictment, the plaintiff amended his complaint, adding an allegation that Northrop was using inadequate damping fluid in its flight data transmitters. The plaintiff admitted that after he left Northrop, he learned of the damping fluid problem through a newspaper article that reported on Northrop’s indictment. The investigation leading up to the indictment may have been based on information that the plaintiff provided to the government. We held that a “disclosure resulting from a criminal investigation by the government based on information provided by a qui tam plaintiff’ does not bar the action under § 3730(e)(4)(A). Id. at 411-12. Biddle asserts that all of the disclosures regarding Stanford’s alleged fraud resulted from a governmental investigation, which was triggered when Biddle provided the government with the pertinent information. Thus, Biddle argues, the disclosures are not “public disclosures” under Barajas.

Stanford contends that this court’s decision in United States ex rel. Devlin v. California, 84 F.3d 358 (9th Cir.), cert. denied, — U.S. -, 117 S.Ct. 361 , 136 L.Ed.2d 252 (1996), is controlling. In Devlin, a government employee informed the plaintiffs of alleged fraud in which the employee had participated by falsifying records. Shortly thereafter, one of the plaintiffs met a newspaper reporter and told him about the alleged fraud. The newspaper then published an article describing the allegedly fraudulent conduct. Five days after the article was published, the plaintiffs filed their qui tam action. This court held that because the action was filed after the newspaper publicly disclosed the allegations of fraud, the district court lacked subject matter jurisdiction, unless the plaintiffs were an “original source” of the information upon which the allegations were based. Id. at 359.

We believe that Devlin is dispositive, and that Biddle misreads Barajas. In Barajas, the allegations against the defendant became public because the government obtained an indictment against the defendant. The court reasoned that if the allegations in the indictment were deemed publicly disclosed, then the government could “limit the potential recovery of qui tam plaintiffs unfairly simply by initiating a criminal investigation.” Barajas, 5 F.3d at 411 . In contrast, Biddle himself was responsible for the story reaching the public in that he disclosed the alleged fraud to the media. For example, Biddle appeared on the news program “20/20,” which aired in March of 1991. Thus, the present case is similar to Devlin, in that the relator in each ease personally disclosed information to the media. The rationale of Barajas that the government should not be able to stifle a plaintiffs qui tam suit by launching a public investigation does not apply where the relator himself was involved in the public disclosure of the allegations. Therefore, we hold that the media reports of Stanford’s alleged fraud were “public disclosures” within the meaning of the FCA.

II. Is Biddle’s action “based upon” public disclosures?

Even though the allegations of Stanford’s fraud were publicly disclosed, the district court would not be without jurisdiction unless Biddle’s action was “based upon” those public disclosures. Biddle contends that “based upon” means “derived from.” Thus, Biddle argues that because he learned first-hand of Stanford’s alleged fraud, rather than through the media reports, his action is not based upon the public disclosures.

Biddle relies on a Fourth Circuit case, United States ex rel. Siller v. Becton Dickinson & Co., 21 F.3d 1339 (4th Cir.), cert. denied, 513 U.S. 928 , 115 S.Ct. 316 , 130 L.Ed.2d 278 (1994). In Siller, the relator was employed by the defendant’s distributor. The distributor filed a suit against the defendant, alleging that the defendant wrongfully terminated its distributorship because the distributor planned to inform the government that the defendant was overcharging the government. Subsequently, the relator filed a qui tam suit against the defendant. Although the allegations of overcharging had been publicly disclosed by the distributor’s complaint, the relator argued that his action was not based upon the public disclosures because his complaint was not derived from those disclosures, but rather, from his own personal experience. The Fourth Circuit agreed, holding that “based upon” means “derived from.” Id. at 1348.

The Fourth Circuit reasoned that the dictionary definition of based upon is to “use as a basis for,” which is synonymous with “derived from.” Id. The Fourth Circuit further reasoned that this construction would still promote the policy of preventing “parasitic” suits because it would disallow a qui tam plaintiff from bringing a lawsuit where the allegations were derived from publicly disclosed information. Id.

In contrast, Stanford argues that “based upon” means “same as” or “supported by.” One of the primary eases upon which Stan ford relies is United States ex rel. Doe v. John Doe Corp., 960 F.2d 318 (2d Cir.1992). In Doe, a corporation was being investigated by the government for alleged fraudulent billing practices. The employee that maintained the false records became a government informant. The informant assigned to his attorney his rights to bring a qui tam action. When the attorney filed his qui tam suit, the allegations of fraud had already been publicly disclosed. In holding that the attorney’s claim was based upon the disclosures, the Second Circuit stated: “The allegations in his complaint are the same as those that had been publicly disclosed prior to the filing of the qui tam suit. Public disclosure of the allegations divests district courts of jurisdiction over qui tam suits, regardless of where the relator obtained his information.” Id. at 324 . The Second Circuit reasoned that a broader construction of “based upon” would result in a flourishing of parasitic suits. Id.

Although this circuit has yet to explicitly address the interpretation of “based upon” in the current version of the FCA, our decisions implicitly support Stanford’s interpretation. First, in Wang v. FMC Corp., 975 F.2d 1412 (9th Cir.1992), the relator brought a qui tam suit against his former employer, alleging that the employer’s performance of various defense contracts was defective. Before the relator filed his complaint, newspaper articles described the defective nature of the defense contracts. Although the district court ruled that the relator’s suit was barred because he was not an “original source” of the information, it would not have had to reach the original source issue unless it had already found that the relator’s claim was based upon public disclosures of the allegations. We stated:

Id. at 1417 (citations omitted). This language supports Stanford’s view that a claim is “based upon” public disclosure when the claim repeats allegations that have already been disclosed to the public.

In United States ex rel. Fine v. Chevron, 72 F.3d 740 (9th Cir.1995) (en banc), cert. denied, 517 U.S. 1233 , 116 S.Ct. 1877 , 135 L.Ed.2d 173 (1996), the relator supervised and edited audits performed by employees under his supervision. After leaving his job, the relator filed a qui tam suit against Chevron alleging fraud against the government. He conceded that he based his claims on reports he furnished to his supervisors, the contents of which were publicly disclosed pursuant to the Inspector General Act. The parties agreed that the allegations satisfied the statutory definition of “based upon public disclosures.” Id. at 741. Although Fine did not directly address the public disclosure issue, it would have been a different case if “based upon” meant “derived from.” The relator’s suit was not derived from public disclosures, but rather, from his first-hand knowledge. Yet his suit was “based upon” public disclosures.

Finally, in Pettis ex rel. United States v. Morrison-Knudsen Co., 577 F.2d 668 (9th Cir.1978), this court interpreted 31 U.S.C. § 232 (C), the predecessor to § 3730(e)(4)(A), which provided that the court “shall have no jurisdiction to proceed with any such suit ... whenever it shall appear that such suit was based upon evidence or information in the possession of the United States ... at the time such suit was brought.” Id. at 670-71. The relator was an engineer on a project financed by the United States. After observing practices he thought to be fraudulent, the relator informed government officials, which led to two audits of the project. The relator also informed a senator, which resulted in an investigation by the General Accounting Of fice (“GAO”). Two weeks after the GAO released its report, the relator filed his qui tam action. This court upheld the district court’s ruling that the relator’s suit was barred because it was based upon evidence already known to the government. Id. at 673. Thus, although the relator did not derive his suit from evidence in the possession of the government, his claim was based upon such evidence.

When Congress amended the FCA in 1986, the jurisdictional bar was changed from one precluding actions “based upon evidence or information in the possession of the United States” to one precluding actions “based upon the public disclosure of allegations or transactions.” However, there is no evidence that Congress intended to change the meaning of “based upon.” See United States ex rel. Findley v. FPC-Boron Employees’ Club, 105 F.3d 675, 684-85 (D.C.Cir.), cert. denied, — U.S. -, 118 S.Ct. 172 , 139 L.Ed.2d 114 (1997). Thus, the Pettis interpretation of “based upon” remains valid under the current statute.

We differ with the Fourth Circuit’s interpretation for several reasons. First, it renders the “original source” requirement largely superfluous. See Findley, 105 F.3d at 683 . If a relator’s lawsuit is based upon public disclosure, then the district court lacks jurisdiction unless the relator is an original source. To be an original source, the relator must have “direct and independent knowledge of the information on which the allegations are based” and must “voluntarily provide[ ] the information to the Government before filing an action.” 31 U.S.C. § 3730 (e)(4)(B).

Robert Salcido, Screening Out Unworthy Whistleblower Actions: An Historical Analysis of the Jurisdictional Bar to Qui Tam Actions, 24 Pub. Cont. L.J. 237 , 273 (1995), cited in Findley, 105 F.3d at 683 . In other words, to say that a relator's complaint is not derived from public disclosures is to say that the relator had direct and independent knowledge of the fraud. Thus, under the Siller view, the “based upon” language in § 3730(e)(4)(A) duplicates the “direct and independent knowledge” language in § 3730(e)(4)(B), allowing the relator to avoid the voluntariness requirement in § 3730(e)(4)(B).

Furthermore, the Siller interpretation would not further the policies underlying the FCA. Two of the primary purposes of the FCA are to alert the government as early as possible to fraud that is being committed against it and to encourage insiders to come forward with such information where they would otherwise have little incentive to do so. See H.R.Rep. No. 99-660, at 22 (1986) (“The purpose of the qui tam provisions of the False Claims Act is to encourage private individuals who are aware of fraud being perpetrated against the Government to bring such information forward.”); S.Rep. No. 99-345, at 14 (1986), reprinted in 1986 U.S.C.C.A.N. 5266, 5279 (stating that the False Claims Act “reward[s] those private individuals who take significant personal risks to bring such wrongdoing to light”); id. at 6, reprinted in 1986 U.S.C.C.A.N. 5266, 5271 (“The Committee believes changes are necessary to halt the so called ‘conspiracy of silence’ that has allowed fraud against the Government to flourish.”); United States ex rel. Dick v. Long Island Lighting Co., 912 F.2d 13, 18 (2d Cir.1990) (holding that the public disclosure limitation “discourages persons with relevant information from remaining silent and encourages them to report such information at the earliest possible time”). These purposes are accomplished by giving the relator a financial incentive to disclose evidence of fraud. However, where neither of these purposes is implicated in a particular case, the relator should not be so rewarded.

Where the relator brings new information of fraud to the government, the relator should be rewarded regardless of how the relator came into possession of that information. The reason is that the allegations in the complaint, being previously undisclosed, are valuable to the government in remedying the fraud that is being committed against it. On the other hand, where the allegations of the fraud are already public knowledge, the relator confers no additional benefit upon the government by subsequently repeating the fraud allegations in his complaint. The relator should be rewarded only if he was an original source of the information. Insiders who have no duty to disclose fraud, i.e., who do so voluntarily, should be encouraged to take “significant personal risks to bring such wrongdoing to light,” and should be rewarded for doing so.

If we were to accept Siller’s interpretation of “based upon” to mean “derived from,” Biddle would be permitted to maintain his lawsuit without ever being required to show that he was the original source of the information, for his complaint was not “derived from” the public disclosures. Assuming, without deciding, that Biddle had an obligation to uncover fraud at Stanford, his disclosure of the fraud would not have been voluntary. In such a case, neither of the above two purposes — to provide the government with new information and to encourage persons to come forward who otherwise would not do so — would be furthered by allowing Biddle to bring his qui tam lawsuit.

This interpretation is consistent with the statutory structure of the FCA. Under § 3730(e)(4)(A), if the allegations were not previously disclosed to the public, the relator’s complaint benefits the government, and the relator is rewarded without inquiring into the details of how the relator obtained the information. If, on the other hand, the allegations in the complaint do not benefit the government because the government already knew about them, then § 3730(e)(4)(A) bars jurisdiction unless the second policy is furthered, that is, an insider provided information to the government who was under no duty to do so.

This interpretation of the FCA is also supported by legislative history.

132 Cong. Rec. H9382 (1986) (statement of Rep. Berman) (emphasis added). Furthermore, every circuit court to address this issue, except for the Fourth Circuit in Siller, has held similarly. See United States ex rel. McKenzie v. BellSouth Telecommunications, Inc., 123 F.3d 935, 940 (6th Cir.1997), cert. denied, — U.S. -, 118 S.Ct. 855 , 139 L.Ed.2d 755 (1998); United States ex rel. Findley v. FPC-Boron Employees’ Club, 105 F.3d 675, 682 (D.C.Cir.), cert. denied, — U.S. -, 118 S.Ct. 172 , 139 L.Ed.2d 114 (1997); United States ex rel. Fine v. MK-Ferguson Co., 99 F.3d 1538, 1546 (10th Cir.1996); Federal Recovery Services, Inc. v. United States, 72 F.3d 447, 451 (5th Cir.1995); Cooper v. Blue Cross & Blue Shield of Florida, 19 F.3d 562, 567 (11th Cir.1994); United States ex rel. Doe v. John Doe Corp., 960 F.2d 318, 324 (2d Cir.1992).

We, therefore, hold that if at the time a relator files a qui tam complaint, the allegations or transactions of the complaint have been publicly disclosed, then the allegations are “based upon” the publicly disclosed information, and the relator must show that he is an original source of the information in order for a district court to have jurisdiction over the lawsuit.

Biddle brought his qui tam suit after the allegations of Stanford’s fraud were publicly disclosed. Thus, the allegations in Biddle’s complaint were “based upon” public disclosures. They did not aid the government, for the government was already aware of the alleged fraud. Accordingly, Biddle should not be rewarded with a potential percentage recovery in a qui tam suit unless he is the original source of the information, i.e., he had direct and independent knowledge of the fraud and he provided the information voluntarily. We now turn to that question.

III. Is Biddle an “original source” under the FCA?

An “original source” is “an individual who has direct and independent knowledge of the information on which the allegations are based and has voluntarily provided the information to the Government before filing an action under this section which is based on the information.” 81 U.S.C. § 3730 (e)(4)(B). Regarding the “direct and independent knowledge” prong, the district court found, and the parties do not dispute, that Biddle had such knowledge of Stanford’s alleged fraud and that he was the source of the government’s information. We agree.

The primary disagreement between the parties is whether Biddle came forth voluntarily, or was instead under a duty to expose fraud. The district court found that Biddle did not provide information to the government voluntarily. Athough the question of whether Biddle was a volunteer is one of law to be reviewed de novo, this court must accept the district court’s findings of fact supporting its conclusion of involuntariness unless they are clearly erroneous. Hagood v. Sonoma County Water Agency, 81 F.3d 1465, 1472 (9th Cir.) (‘Hagood II”), cert. denied, — U.S. -, 117 S.Ct. 175 , 136 L.Ed.2d 116 (1996). Biddle bears the burden of establishing jurisdiction by a preponderance of the evidence. Id.

The two primary Ninth Circuit cases to address the issue of voluntariness are Fine and Hagood II. In Fine, the relator was an auditor for the Office of Inspector General. His job required him to supervise audits that other employees had conducted, and edit audit reports that others had written. After leaving his job, the relator filed a qui tam suit against Chevron alleging fraud against the government. This court held that the relator did not disclose the fraud voluntarily because “[h]e was a salaried government employee, compelled to disclose fraud by the very terms of his employment.” Fine, 72 F.3d at 743 .

In Hagood II, the relator was assistant district counsel to the Amy Corps of Engineers. He was assigned to represent the district counsel in handling the renegotiation of a certain contract. The relator disclosed the fact that the contract did not comply with the Water Supply Act. This court held that he voluntarily disclosed the information to the government. In distinguishing Fine, the court stated:

Hagood II, 81 F.3d at 1476 n. 19. But see id. at 1480 (Kleinfeld, J., concurring) (believing that the relator did not come forth voluntarily because “a lawyer working on a transaction has a duty as an agent to disclose to his principal ‘information relevant to matters within his province and of which he should know the principal would want to know”’) (quoting Warren A. Seavey, Law of Agency § 143 (1964)).

The present case involves allegations of fraud arising out of two types of contracts: those negotiated while Biddle was employed, and those negotiated before Biddle was hired.

A. Contracts Negotiated While Biddle Was Employed

As stated above, Biddle was the Resident Representative Administrative Contracting Officer (“ACO”) for the Office of Naval Research at Stanford. The district court noted that Biddle was not an auditor like the relator in Fine whose “paramount responsibility” was to disclose fraud. Rather, auditing is performed by the Defense Contract Audit Agency (“DCAA”). However, the court found that Biddle’s job duties “included some aspects of such work.” In reaching its conclusion, the district court relied in part on Biddle’s job responsibilities as defined in the Federal Acquisition Regulations (“FAR”). Under FAR . § 1.602-1, “[c]ontracting officers have authority to enter into, administer, or terminate contracts and make related determinations and findings.” 48 C.F.R. § 1.602 -l(a). “No contract shall be entered into unless the contracting officer ensures that all requirements of law, executive orders, regulations, and all other applicable procedures, including clearances and approvals, have been met.” Id. § 1.602-1(b). Moreover, “[cjontracting officers are responsible for ensuring performance of all necessary actions for effective contracting, ensuring compliance with the terms of the contract, and safeguarding the interests of the United States in its contractual relationships. ” Id. § 1.602-2 (emphasis added).

Another section of FAR provides: “If the contractor is unable to support any part of the claim and there is evidence that the inability is attributable to misrepresentation of fact or to fraud on the part of the contractor, the contracting officer shall refer the matter to the agency official responsible for investigating fraud. ” Id. § 33.209 (emphasis added); see also Navy Acquisition Procedures Supplement (“NAPS”) § 33.209 (directing ACO to report fraud to Navy Inspector General); 48 C.F.R. § 52.215-5 (permitting ACO to examine and audit eon-tractor’s records under certain circumstances).

Finally, the district court supported its conclusion by referring to a report by the Chief of Naval Research. The report stated:

(Zumwalt Decl. Ex. D (1 — 19—91 “Report of Inquiry into Stanford University Overhead Rates” by Chief of Naval Research) at 11).

Biddle argues that the district court erred because he is no different from the relator in Hagood II. Despite the regulatory descriptions of his duties, Biddle contends that when he brought his concerns to the attention, of his supervisors, he was told: “Remember Paul — you’re not a bean counter now, you’re an ACO. If DCAA does not advise of us problems with these things, it’s not for us to go in and find problems with them.” (Biddle Decl. ¶ 11; Opening Brief at 50). Thus, Biddle contends that his job duties did not entail detection of fraud. He further argues that his disclosure of the alleged fraud was necessarily involuntary in light of the opposition exhibited by his superiors. We disagree.

First, Biddle’s duties as an ACO are defined by the applicable regulations and other official documentation of his position, as well as the oral statements of his supervisors. Biddle’s supervisors may direct him to perform certain specific tasks, but in order to determine his broader responsibilities as an ACO, Biddle must examine the pertinent regulations, such as FAR and NAPS. His documented job responsibilities make it clear that his position entailed more than simply drafting contracts like the relator in Hagood II; Biddle had a duty as an ACO to disclose fraud. As discussed above, FAR and NAPS obligated Biddle to ensure the legitimacy of contracts and to refer possible fraud to the appropriate authorities. If Biddle’s argument were taken to its logical extreme, the relator in Fine, who was an auditor for the Inspector General, could have maintained that he had no duty to uncover fraud if his superiors had obstructed his ability to conduct certain audits. The documented responsibilities of a government employee cannot be so easily ignored.

Nor does the fact that Biddle met resistance by his superiors transform his unearthing of possible fraud into a voluntary investigation. The purposes underlying the FCA are to encourage individuals to report potential fraudulent activities being committed against the government, and to reward them for doing so. But where a government employee is paid to perform this function, the employee should not receive a windfall for merely doing his job. Biddle earned a government salary to perform various tasks, one of which was to report suspected fraud to his superiors. Thus, he was a “salaried government employee, compelled to disclose fraud by the very terms of his employment.” Fine, 72 F.3d at 743 . This was true in Fine, even though Fine’s superiors “could not or would not take action against every perceived violation he brought to their attention.” Id. at 742 . Uncovering fraud is a formidable task, especially when one’s supervisors may be opposed, uninterested, or even involved in the fraud. But in accepting the position as ACO, Biddle assumed the risk of encountering such a difficult situation. Why should he now receive a possible windfall for simply doing that which he was required to do and was paid to do?

Furthermore, as the Fine court recognized, by allowing Biddle potentially to recover a substantial amount of money in a qui tarn action, we would be creating “perverse incentives” for similarly-situated government employees. See id. at 745 . Rather than perform their jobs as they are required, government employees obligated to disclose suspected fraud may inappropriately hide fraud from their supervisors while preparing their qui tato actions for filing. Id. Moreover, by racing the government to the courthouse in order to secure personal financial gain, a relator may prematurely alert the target before the government has sufficient time fully to investigate the alleged fraud, thereby frustrating a future case against the target. Id. These policy concerns further bolster our view that Biddle did not voluntarily provide information through his qui tam complaint.

In determining whether Biddle came forth voluntarily, the district court examined the pertinent regulations and other official documentation of Biddle’s responsibilities as an ACO. The court then concluded:

Upon reviewing the evidence relied upon by the district court, we believe that the court’s above findings are not clearly erroneous. Based on those findings, we hold that Biddle did not voluntarily disclose the alleged fraud regarding the contracts that were negotiated while he was employed as an ACO.

B. Contracts Negotiated Before Biddle Was Hired

Biddle was hired as ACO in October of 1988. His complaint, however, alleges fraud committed by Stanford in contracts dating back to 1981. Biddle argues that even if he had a duty to disclose fraud regarding contracts that were negotiated while he was employed as an ACO, he had no such duty with respect to contracts negotiated before he was hired. Athough the district court stated that this issue presented a “closer question,” it ultimately rejected Biddle’s argument.

The court relied in part on a June 2, 1991 memo by Biddle that stated that his “current taskings” included coordination and assistance to DCAA in the conduct of audits of indirect costs and staff benefits for the years 1981-1991. Biddle argues that his tasks were expanded after he disclosed Stanford’s alleged fraud. Therefore, he asserts that the court should not have considered this document, or any other document defining his duties that was issued after he initially disclosed the alleged fraud. He argues that otherwise the government could stifle qui tam suits by merely enlarging a relator’s job duties after initial disclosure so that further disclosure is involuntary.

Even if Biddle’s argument has merit, there was sufficient evidence apart from this memo to find that Biddle had a duty to disclose fraud relating to any contract. First, under FAR and NAPS, as discussed above, Biddle was required to safeguard the interests of the United States and report fraud to the Navy Inspector General. These requirements do not appear to limit Biddle’s duties to contracts negotiated. while he was employed as an ACO. Furthermore, the district court found that in analyzing Stanford’s cost proposals, negotiating rates, and evaluating compliance during his term as ACO, Biddle had to consider overhead rates for previous years. One of the main allegations in Biddle’s complaint was that Stanford was claiming and obtaining excess reimbursement for indirect costs based on contracts negotiated before Biddle was hired. These contracts were still in effect, and allegedly being abused by Stanford, throughout Biddle’s tenure. Upon discovering possible fraud with respect to these on-going contracts, Biddle was under an obligation to disclose it.

Upon reviewing the evidence relied upon by the district court, we hold that the court’s findings are not clearly erroneous. Based on those findings, we hold that Biddle did not voluntarily disclose the alleged fraud regarding the contracts that were negotiated before he was hired. Accordingly, Biddle was not an “original source” of the information that he disclosed to the government, and the district court lacked subject matter jurisdiction over his qui tam action.

AFFIRMED.

. The court expressly limited its holding to whether the relator was an original source. Id. at 743 . . One could argue that this outcome is harsh because, although Biddle's complaint did not reveal new information to the government, Biddle was responsible for publicly disclosing Stanford's alleged fraud in the first place. It must be remembered, however, that Biddle’s failure to satisfy the "based upon public disclosure” requirement is not the end of the inquiry. If Biddle were deemed an original source of the information, the district court could assert jurisdiction over his qui tam suit. The public disclosure requirement is only the initial inquiry to determine whether the allegations in the relator’s complaint provided the government with any valuable information of which it was not already aware. Insiders who come forth voluntarily, and who deserve to be rewarded for doing so, will not be barred merely because their complaints do not reveal any previously undisclosed information. . Biddle argues that the district court’s reliance on § 33.209 of FAR was misplaced in that it applies only to claims for underpayment of government contracts. (Open. Brief at 47). However, "claim” is defined as "a written demand or written assertion [other than a routine request for payment] by one of the contracting parties seeking, as a matter of right, the payment of money in a sum certain, the adjustment or interpretation of contract terms, or other relief arising under or relating to the contract." 48 C.F.R. § 33.201 ; see also 41 U.S.C. § 605 (1987 & Supp.1997). A "claim” does not have to be in dispute as to liability or amount. D.L. Braughler Co. v. West, 127 F.3d 1476 , 1483 n. 8 (Fed.Cir.1997). . An additional issue brought up by the parties is the effect of Executive Order No. 12,674, which provides, in pertinent part: "Employees shall disclose waste, fraud, abuse, and corruption to appropriate authorities." 54 Fed.Reg. 15,159, § 101 (k) (1989). The question is whether the order prevents a federal employee from being an original source. This issue was raised in Fine, 72 F.3d at 744 , but the court declined to address it because it found that the relator had a specific job duty to disclose fraud as an auditor for the Office of Inspector General. Nor was the question answered in Hagood II. The fraud alleged by the relator in that case occurred prior to 1989, the year the Executive Order became effective. See United States ex rel. Hagood v. Sonoma County Water Agency, 929 F.2d 1416 (9th Cir.1991) ("Hagood I"). Because we hold that the regulations defining Biddle's job responsibilities rendered his disclosure involuntary, we need not reach this question.