The major issue in this appeal is a troublesome question concerning the correct construction of the federal bank fraud statute. We are called upon to construe the breadth of a statute on which the Courts of Appeals are divided, and on which our own court has not spoken definitively. A grand jury in the United States District Court for the Eastern District of Pennsylvania returned a three-count indictment charging the defendant, Lisa Thomas, with two counts of bank fraud in violation of 18 U.S.C. § 1344 and one count of fraudulently inducing a person to travel in interstate commerce in violation of 18 U.S.C. § 2314. The District Court granted the Government's motion to dismiss Count III, one of the two bank fraud counts. A jury found the defendant guilty on the remaining two counts. The defendant's motion for judgment of acquittal was denied. The District Court sentenced the defendant to two concurrent thirty-three month sentences, supervised release, and restitution in the sum of $133,300.

Prior to sentence, the defendant objected to the imposition of a two-level upward adjustment for abuse of trust pursuant to U.S.S.G. § 3B1.3 and requested a two-level downward adjustment for acceptance of responsibility pursuant to U.S.S.G. § 3E1.1. The District Court denied both requests. The defendant timely appealed her convictions of bank and travel fraud and related sentencing issues. We reverse the conviction as to bank fraud and affirm the conviction as to travel fraud, and remand for resentencing.

The primary issue on appeal is whether there was sufficient evidence to sustain Thomas's conviction of bank fraud in violation of 18 U.S.C. § 1344. Anne Weygandt, then aged 88, employed Thomas as a home health care aide in and around 1998. Weygandt believed herself to be in fair health during that period, although she had suffered a small stroke in 1997. Around that time, Weygandt frequently made loans to her nephew and also authorized others, including Thomas, to complete checks which she had pre-signed, by filling in the amount and name of the payee. These checks were used for various purposes, including the payment of bills. Thomas also received and sorted Weygandt's mail. From November 1997 to July 1998, Thomas induced Weygandt to sign numerous checks for the pretextual purpose of transferring money among Weygandt's several bank accounts or for the purchase of groceries. Instead, Thomas cashed the checks, made out either to Thomas or to cash, at Weygandt's banks, and pocketed all or most of the proceeds. She withdrew approximately $124,300 from Weygandt's Mellon Bank accounts and $9,400 from her Citizen's Bank account.

Weygandt was physically present at the bank with Thomas when the withdrawals occurred, and she herself endorsed those checks made out to cash. After Thomas originally sought to cash Weygandt's checks by herself, one of the tellers insisted that Weygandt be present before the bank would honor the checks. Despite Weygandt's presence, the transactions still aroused the suspicion of bank tellers, who asked Thomas the purpose of the withdrawals. Either Thomas or Weygandt would always respond that the money was for travel, or for transfers among Weygandt's accounts, or for shopping. A teller showed Weygandt her account balance on at least one occasion, to be sure she grasped the magnitude of her withdrawals. Notwithstanding, Weygandt had no idea of the amounts being withdrawn, or their true purpose. Weygandt physically received the money from the teller some of the time, and on other occasions, Thomas received the money. However, Weygandt repeatedly expressed her authorization of the withdrawals when the tellers inquired, and never repudiated the transactions. Despite suspicions over the validity of the withdrawals, given their frequency and the amount of cash being issued, bank staff never communicated with police or their internal fraud investigators.

Weygandt's nephew became apprehensive of Thomas's conduct and communicated with the police. A State Police investigator confronted Thomas, and she later admitted in a written statement that Weygandt requested her assistance in writing her checks to pay bills, because Weygandt could not fully write them out herself. Thomas went on to state that, because she needed money to fund her drug addiction, she convinced Weygandt to sign checks for her on the pretense of transferring money among her bank accounts, Weygandt having asked her to transfer money for legitimate purposes in the past, and thus being unlikely to become suspicious.

At trial, defense counsel argued essentially that the facts here do not constitute a federal crime of bank fraud. It was not seriously contested that Thomas had acted wrongfully. However, the defense contended that the federal bank fraud statute required that the defendant intend to cause the bank a loss and that the defendant make a material misrepresentation to the bank. Here, the defense argued, the banks were not exposed to a loss as a result of honoring Weygandt's checks, because the checks were properly made payable to Thomas or to cash, and Weygandt had vouched for their legitimacy. Thus, only Weygandt suffered losses and the banks were not subject to any losses or potential liability for honoring the checks. Furthermore, Thomas contended that there was no material misrepresentation because Thomas had not affirmatively deceived the bank, but had merely presented the checks and passively accepted the proceeds.

At trial, Thomas also objected to the admission of a handwritten summary by a State Police investigator listing all the checks cashed by Thomas and the monies converted. At the end of the 6-page list, itemizing each individual check, appeared the statement: "Total Value of Fraud from Mellon Checking $118,550.00." Thomas asserted that the word "Fraud" should have been redacted. She also argued that the District Court's curative instruction, informing the jury that fraud was a conclusion for it to make, not the witness, was insufficient to overcome the resulting prejudice.

We have jurisdiction pursuant to 28 U.S.C. § 1291 over a judgment of conviction and sentence. Our review of a district court's interpretation of the scope and coverage of the bank fraud statute is plenary. United States v. Schwartz, 899 F.2d 243, 243 n. 1 (3d Cir.1990).

The federal bank fraud statute briefly provides:

Whoever knowingly executes, or attempts to execute, a scheme or artifice —

(1) to defraud a financial institution; or

(2) to obtain any of the moneys, funds, credits, assets, securities, or other property owned by, or under the custody or control of, a financial institution, by means of false or fraudulent pretenses, representations, or promises;

shall be fined not more than $1,000,000 or imprisoned not more than 30 years, or both.

18 U.S.C. § 1344.

The meaning of the first line of the statute is not disputed. "The terms `scheme' and `artifice' are defined to include any plan, pattern or cause of action, including false and fraudulent pretenses and misrepresentations, intended to deceive others in order to obtain something of value, such as money, from the institution to be deceived." United States v. Goldblatt, 813 F.2d 619, 624 (3d Cir.1987). As to subsections (1) and (2), the Government maintains:

Both subsections prohibit schemes or artifices fraudulently to obtain money or property owned by or held in the custody of a financial institution. The difference is that, under subsection (1), the fraud victim must be a bank, whereas under subsection (2), the victim need not be a bank as long as property under the custody and control of a bank is obtained through false and fraudulent pretenses or representations.

Government's brief at 27. The Government also notes that the indictment charged Thomas with both prongs of the statute which permitted Thomas to be convicted if the Government proved the elements of either subsection. It contends that our decision in United States v. Monostra, 125 F.3d 183 (3d Cir.1997), holds that the requisite criminal intent "may be met by the government showing that the defendant engaged in conduct with a financial institution which resulted in the improper release of money deposited with the institution," or by showing that the bank was exposed to a loss of its own property. Government's brief at 29, 30-31.

As Thomas admits in her confession, her crime involved a pattern of activity intended to deceive others, including acquiring Weygandt's trust, making deceptive misrepresentations to her, and some to the bank. The deceptions were employed systematically by Thomas and constituted a manifest departure from fundamental honesty. The issue before us, however, is not whether there was a scheme or artifice afoot; rather, we must address whether that scheme defrauded or attempted to defraud a financial institution in violation of the statute.

Subsection (1) requires that the scheme or artifice must be intended "to defraud a financial institution." We have held that a "scheme to defraud" is measured "by determining whether the scheme demonstrated a departure from fundamental honesty, moral uprightness, or fair play and candid dealings in the general life of the community." Goldblatt, 813 F.2d at 624. The statute's use of the term thus means that the defendant, through the exercise of a scheme that departs from fundamental honesty, must thereby intend to defraud the bank.

Subsection (2), however, facially requires only that the perpetrator engage in a "scheme or artifice" in order to obtain bank funds or funds in bank custody. The use of the disjunctive "or" connecting the two subsections seems to indicate that the two connected subsections of the statute are to be given independent, or disjunctive effect. This is also the Government's position. It asserts that, under subsection (2), the victim "need not be a bank as long as property under the custody and control of a bank is obtained." A disjunctive reading of the two sections, as proposed by the Government, gives the statute a breadth of scope that extends well beyond what Congress intended the statute to regulate. Subsection (2), unlike (1), provides only the most tenuous nexus between the scheme or artifice and the institution of banking, which Congress sought foremost to protect. An examination of the Congressional history of the statute reveals that Congress enacted the statute for the purpose of protecting financial institutions from the perpetration of fraud on them, leaving to states the traditional prosecution of crimes of larceny, embezzlement and fraudulent conversions. See S.Rep. No. 98-225 at 377 (1984), reprinted in 1984 U.S.C.C.A.N. 3182, 3517 ("Clearly there is a strong federal interest in protecting the financial integrity of [banking] institutions.").

Under the Government's theory, almost any scheme in which a victim withdraws money from a bank and turns it over to the perpetrator would become fair game under the statute. Such a reading of the statute is irreconcilable with Congressional intent; such conduct has only a remote and hypothetical effect on the integrity of banking. Subsection (1), which requires a nexus of harm or loss to the bank, seems a far more rational expression of the federal interest here. Nonetheless, the "plain meaning" is our starting point. We do not lightly disregard the statutory language. Immigration and Naturalization Serv. v. Elias-Zacarias, 502 U.S. 478, 482, 112 S. Ct. 812, 117 L. Ed. 2d 38 (1992). However the weight of the Congressional legislative history and the plausible construction of it by some of our sister courts counsel in favor of a conjunctive reading of the subsections of the statute, limiting it to fraud perpetrated on financial institutions.

The Courts of Appeals are not of one mind as to the proper reading of the statute, including whether the intent requirement of subsection (1) applies to any indictment pled under the statute, or whether subsection (2) can be read wholly independently of subsection (1). See United States v. Everett, 270 F.3d 986, 990 (6th Cir.2001). This court has spoken equivocally on the matter, making it extremely difficult for the District Court to discern the applicable state of the law. We have left an open question whether it is appropriate to read subsection (2) disjunctively. Despite the court's observation in Monostra that there is substantial evidence that Congress intended subsection (2) to underscore the scope of subsection (1), rather than to set forth a separate offense, it merely suggested, without making a definitive ruling, that the two subsections are to be read in unison. See Monostra 125 F.3d at 183. In an earlier decision, this court held that the two subsections were to be read disjunctively. See Schwartz, 899 F.2d at 248.

We note, however, that in Schwartz the court considered only whether an indictment pled solely under subsection (1) must allege "false or fraudulent pretenses, representations or promises," factors which are stated only in subsection (2). The court held that it need not, and that a case pled under subsection (1) does not necessarily require that any element of subsection (2) must also be proven. Id. at 246. The later Monostra decision did not hold that subsection (2) limits the scope of subsection (1), but rather that subsection (2) broadens the scope of subsection (1) to include situations where property merely in "the custody of" the bank is taken. In Monostra, the court considered whether a prosecution under subsection (1) is deficient because it involves the taking of a depositor's funds, rather than a bank's funds, and because subsection (1) on its face does not cover such instances. 125 F.3d at 186. Monostra held that the two subsections should probably be read conjunctively to the extent that subsection (2) underscores the range of situations in which the statute might apply, including the taking of funds in bank custody. Monostra did not hold that subsection (1), read alone in the disjunctive, could not itself set forth a bank fraud violation. To the contrary, its holding contemplates that subsection (1) on its own terms does set forth a violation and that subsection (2) simply broadens its scope. Id. at 187. ("Given Congress's aim of creating a statute that would empower federal prosecutors to pursue all forms of bank fraud, it is evident that § 1344(2) was mainly intended to underscore the breadth of the statute's reach.") This merely clarifies Schwartz's holding.

What Monostra did not hold, but what its reasoning plainly suggests, is that there can be no such thing as an independent violation under subsection (2). To convict at all under the bank fraud statute, there must be an intent to defraud the bank. Bank fraud may involve a scheme to take a bank's own funds, or it may involve a scheme to take funds merely in a bank's custody. Similarly, it may involve a scheme involving "false or fraudulent pretenses, representations or promises." We recognize that although "false or fraudulent pretenses, representations or promises" are optional under the statute, a material representation is a required element of proof to show any violation of the bank fraud statute. Neder v. United States, 527 U.S. 1, 25, 119 S. Ct. 1827, 144 L. Ed. 2d 35 (1999) ("[W]e hold that materiality of falsehood is an element of the federal mail fraud, wire fraud, and bank fraud statutes.") But the sine qua non of a bank fraud violation, no matter what subdivision of the statute it is pled under, is the intent to defraud the bank.

To reach this conclusion, we have plumbed the Congressional history. Congress enacted the bank fraud statute to fill the gaps existing in federal jurisdiction over "frauds in which the victims are financial institutions that are federally created, controlled or insured." S.Rep. No. 98-225 at 377 (1984), reprinted in 1984 U.S.C.C.A.N. 3182, 3517. The statute is primarily concerned with "fraudulent schemes where banks are victims." H.R.Rep. No. 98-901 (1984). These pronouncements in the legislative history strongly suggest that the legislature wanted the intent requirements of subsection (1) to apply to any indictment under the statute, and that, in order to prove bank fraud, a bank must be more than a mere incidental player. A defendant must have deliberately targeted his or her scheme at the banking institution.

Moreover, as Judge Nygaard noted in Monostra, Congress modeled the bank fraud statute closely upon the mail fraud statute, which has very similar language. 125 F.3d at 186-87; see 18 U.S.C. § 1341. Courts have routinely looked to Congressional intent with respect to construction of the mail fraud statute for guidance in interpreting the bank fraud statute. 125 F.3d at 186-87. In Monostra, we noted that the Supreme Court, in McNally v. United States, 483 U.S. 350, 107 S. Ct. 2875, 97 L. Ed. 2d 292 (1987), held that Congress intended the second dependent clause of the mail fraud statute to broaden the scope of the first clause. The mail fraud statute was thus intended to cover "any scheme or artifice to defraud," including any "[scheme] for obtaining money or property by means of false or fraudulent... promises." Id. at 351, 107 S. Ct. 2875. According to McNally, the second clause made clear that the statute extended to cover deceptive misrepresentations about the future, i.e., promises, as well as misrepresentations about the present. Because the bank fraud statute was modeled after the mail fraud statute, the correct syntactical construction of the mail fraud statute sheds light on the appropriate construction of the bank fraud statute. Thus, subsection (2) does not set forth an independent basis of liability.

Read in this light, subsection (2) of the bank fraud statute "underscores the breadth" of subsection (1). Under Monostra, subsection (2) would not provide a separate basis of criminal liability under the statute. The Court of Appeals for the Second Circuit in United States v. Blackmon, 839 F.2d 900 (2d Cir.1988), largely concurs in this assessment. In Blackmon, the court noted that the requisite criminal intent to "victimize a bank" which is apparent from reading the Senate reports on the statute, and which the courts have implied into subsection (1), was intended to apply to all crimes alleged under the statute. Id. at 905-06. Thus, to take money in the custody of a bank is not a crime under the statute unless there is a concomitant intent to victimize the bank. In United States v. Rodriguez, 140 F.3d 163, 167 n. 2 (2d Cir.1998), the Court of Appeals again held that a deceptive pattern of conduct designed to deceive a bank was required to prove a case under either subsection (1) or (2).1 The Courts of Appeals for the Fifth and Seventh Circuits have also determined that the intent-to-victimize requirement of subsection (1) pervades the statute, and is a necessary element of an indictment under either subsection (1) or (2). Accord, United States v. Sprick, 233 F.3d 845, 852 (5th Cir.2000); United States v. Davis, 989 F.2d 244, 246-47 (7th Cir.1993).2

The Second Circuit Court of Appeals rejected a stark reading of subsection (2) in Blackmon, a pigeon drop case. In a pigeon drop scheme, the victim is induced to take money out of the bank and to hand that money over to the perpetrators of the scheme. Id. at 903. In some attenuated sense, the perpetrators did intentionally cause the loss of funds in bank custody, inasmuch as they knew the money was withdrawn solely for the purposes of the victim's participation in their scheme. The court, however, rejected the proposition that this crime should fall under the bank fraud statute, because there was no way the crime could have been viewed as intended to victimize or defraud the bank. Id. at 905. Although the funds were derived from the bank, it could not reasonably be said that the fraud harmed the bank's integrity. Id. at 906 (noting that protecting a bank's integrity underlies the Congressional intent). Money is taken from banks every day for countless foolish purposes, but in such instances, banks are not exposed to liability nor is their integrity compromised. Moreover, to hold otherwise would seriously diminish the jurisdiction of state criminal law.

Thus, in Blackmon, the court noted that "terms connected in the disjunctive need not always be construed independently so that the limits applicable to one term are inapplicable to the second, especially when such a construction would leave the statute's outer boundaries ambiguous, and involve the federal government in areas more properly left to states and localities." Id. at 905 n. 5. Thus the court held that subsection (2) is delimited by the intent requirements of subsection (1) of the statute. Absent a limiting principle, the outer reach of the statute is extended far beyond what Congress intended.

In United States v. Bass, 404 U.S. 336, 349, 92 S. Ct. 515, 30 L. Ed. 2d 488 (1971), the Supreme Court held that "unless Congress conveys its purpose clearly, it will not be deemed to have significantly changed the federal-state balance. Congress has traditionally been reluctant to define as a federal crime conduct readily denounced as criminal by the States." We are disinclined to read a statute in a manner which permits it to trench on the domain of traditional state criminal law, circumscribed only by the exercise of prosecutorial discretion. The extension proposed here by the Government offends the balance of federal and state jurisdiction and our principles of comity by imposing federal law where the federal interest is remote and attenuated. We hold, therefore, that conduct, reprehensible as it may be, does not fall within the ambit of the bank fraud statute when the intention of the wrongdoer is not to defraud or expose the bank to any loss but solely to defraud the bank's customer. Our holding today is consonant, not only with Congressional intent but also with our federal criminal law jurisprudence.

Our holding that the statute is to be read conjunctively does not end this matter. We must still decide the thorny question of what is meant by the subsection (1) requirement that the defendant intends to defraud the bank. The Government's position is that the banks were exposed "to the real threat of civil liability for having succumbed to Thomas's conduct and the loss of the use of the money that was withdrawn." The Government, however, cites no authority for this proposition, one that is highly speculative. We see no evidence that there was civil liability or that Thomas intended to expose the bank to a loss. The Government also suggests that mere "deceptive conduct" toward the bank establishes intent to defraud. We disagree. See, e.g., State v. Weigel, 194 N.J.Super. 451, 477 A.2d 372, 194 N.J.Super. 451, 462 (1984) ("In common parlance, the word `defraud' means to cheat or wrongfully deprive another of his property by deception or artifice.") Congress sought to proscribe conduct that "victimize[d]" banks, which suggests that the bank must be deliberately harmed before the statute is violated. We believe that, given the legislative intent, harm or loss to the bank must be contemplated by the wrongdoer to make out a crime of bank fraud. See United States v. Brandon, 298 F.3d 307, 312 (4th Cir.2002); United States v. Sprick, 233 F.3d 845, 852 (5th Cir.2000); United States v. Blackmon, 839 F.2d at 905-06; United States v. Moede, 48 F.3d 238, 242 (7th Cir.1995).3

The legislative history shows that Congress sought to allay crimes that undermined public confidence in banking institutions. See Blackmon, 839 F.2d at 906. ("[W]here the victim is not a bank and the fraud does not threaten the financial integrity of a federally controlled or insured bank, there seems to be no basis in the legislative history for finding coverage under [subsection (2)].") The deception of a bank as an incidental part of a scheme primarily intended to bilk a bank customer does not undermine the integrity of banking. Our reading of the Congressional intent is also supported by the Court of Appeals of the Seventh Circuit. Writing for that court, Judge Posner stated that "the purpose [of the bank fraud statute] is not to protect people who write checks to con artists but to protect the federal government's interest as an insurer of financial institutions." Davis, 989 F.2d at 247. In that case, facially valid checks, issued by the IRS as a tax refund to a person who, in fact, was owed no tax refund, were deposited at a bank. The bank was, of course, deceived as to the entitlement of the payee to the money, whether by affirmative deception or passive, material omission. Yet, according to Judge Posner, only when the Government proves the existence of civil liability is the bank truly endangered by the fraud. Cashing facially valid checks, even where the bank is deceived as to their legitimacy, does not expose the bank to liability and, therefore, there is no bank fraud. Honoring those checks did not "endanger" the bank's integrity. Id. at 247. People write checks "to con artists" or for many other imprudent purposes every day, but this does not undermine the integrity of the bank as to give rise to a federal interest. Clearly, however, when the scheme targets the bank as its direct victim, not a depositor, and the bank suffers a loss of its funds, there is no civil liability requirement. This was the case in United States v. Goldblatt, 813 F.2d 619, 623-24 (3d Cir.1987).

The Court of Appeals for the Second Circuit also holds, and we agree, that a defendant must intend to cause a bank a loss or potential liability, whether by way of "statutory law, common law, or business practice." United States v. Laljie, 184 F.3d 180, 191 (2d Cir.1999). We believe that this holding is effective shorthand for the legislative intent that the statute would proscribe conduct which undermines the integrity of federally insured institutions. See S.Rep. No. 98-225, at 377 (1984), reprinted in 1984 U.S.C.C.A.N. 3182, 3517 (noting that the statute was intended to protect the "financial integrity" of banks). The reputation and integrity of banking are harmed when a bank is victimized in a way that exposes it to liability. Conduct which exposes a bank to liability casts the institution of banking into doubt by adversely affecting its image with the public. It implicates the federal interest in maintaining the integrity and esteem of our federally insured banks.

Laljie illustrates the kind of distinction we make between schemes which victimize banks by exposing them to liability or loss, and schemes in which banks, despite being the target of deception, are mere "unwitting instrumentalities" to the fraud. 184 F.3d at 190. Laljie addressed two distinct accusations of bank fraud. The first involved an employee's presentation of altered employer checks to the bank. The defendant's actions were directed at deceiving the bank and exposed the bank to civil liability for honoring the checks under New York law, under which a bank may be held liable for paying conspicuously forged or altered checks. In the latter scenario, however, the employer signed the check, but deliberately left the payee line blank. The defendant filled in the payee line but did not otherwise alter the check. Although he deceived the bank, he did not expose it to any loss. His conduct solely victimized the maker of the check, and not the bank, which incurred no loss. The bank merely served as "the unwitting instrumentality of the fraud."

The victim in this latter case entrusted the defendant with signed checks with the payees left blank. There is no meaningful difference between this situation and one in which an employer gave an employee keys to its cash box. On the other hand, when an employee cashes some employer's forged checks, he or she harms the bank. The public expects that a properly completed check will not be used by other than its intended beneficiary and that banks ought to be vigilant of forged or altered checks, and civil liability reflects this public expectation. "The purpose of the Commercial Code is to enhance the marketability of negotiable instruments and to allow bankers, brokers, and the general public to trade in confidence." Manor Bldg. Corp. v. Manor Complex Associates, Ltd., 435 Pa.Super. 246, 645 A.2d 843, 846 (1994). The law governing bank liability for paying on a fraudulent negotiable instrument thus serves to vouchsafe public confidence in the banking system. However, the public is unlikely to lose confidence in a financial institution because it honors checks that some individuals foolishly have allowed others to complete.

Returning to the case at hand, we examine Delaware law concerning the loss effect of Thomas's fraudulent endorsement or misuse of an employer's checks, because the two banks on which the victim's funds were drawn were in Delaware. The Government does not raise seriously the issue of the banks' loss but assumes that the mere act of presenting fraudulently drawn checks to a bank for payment constitutes a loss to it or a potential for liability. The Government has not pointed to any case that establishes actual or bank potential liability in such situations, nor do we perceive any. Delaware law, in this area, reflects the belief that the depositor is in a better position to avoid the loss by carefully selecting its employees or agents, in supervising them, and in adopting measures to prevent forged endorsements on the instruments drawn against the depositor. Thus, a bank is subject to liability only when it fails to utilize "ordinary care in paying or taking the instrument" when it is presented. DEL. CODE ANN. TIT. 6 § 3-405. So long as a bank utilizes due care, it may not be penalized for what is essentially the negligence of its depositor.4

Where, however, an employee or agent is authorized by a depositor to endorse the depositor's checks, the appropriate Uniform Commercial Code provision is DEL. CODE ANN. TIT. 6 § 3-307. It provides that where a check is endorsed by a person so empowered and "made payable to the[endorsee] personally, the taker does not have notice of the breach of fiduciary duty unless the taker knows of the breach of fiduciary duty." The depositor may have owed the endorsee money for any number of legitimate reasons, and a bank cannot be held liable for a transaction which appears facially sound.

In this case, Weygandt authorized Thomas to fill out the payee line of the checks and the amount. The checks were made out to cash or to Thomas personally. The bank has no liability under section 3-405 in such circumstances, inasmuch as there are no forged endorsements at issue. Pursuant to section 3-307, a bank cannot be held liable for abetting Thomas's breach of fiduciary duty when the checks were made out to Thomas personally, or to cash, unless it actually knew of the breach of duty. A bank is not "responsible for knowing to what entities the owner of the account might make payments." Laljie, 184 F.3d at 191 (discussing New York Uniform Commercial Code).

Moreover, even were there a colorable case for civil liability set forth here, it must also be shown that Thomas intended to victimize the bank. Even a scheme which does expose a bank to a loss must be so intended. "[A] scheme to pass bad checks [to merchants] is not bank fraud," because, even though the bank might honor the checks and be civilly liable, the defendant did not anticipate that the bank, rather than the merchant, would bear the loss. United States v. Jacobs, 117 F.3d 82, 93 (2d Cir.1997). In United States v. Barrett, 178 F.3d 643, 648 (2d Cir.1999) the court held that one must look to the "entire circumstances of defendant's conduct as an indication of the requisite criminal intent." Thomas's actions, in fact, demonstrate that she never intended to victimize the banks. Her only victim was Weygandt.

Thomas argues that the admission of documents prepared by the investigating police officer, King, was reversible error. In a summary of all the checks cashed by Thomas, King wrote, "Total Value of Fraud from Mellon Checking $118550.00." Thomas argues that the existence of fraud was a conclusion to be made by the jury and that the use of the word "fraud" by a Government witness was a usurpation of this jury function.

In United States v. Zehrbach, 47 F.3d 1252 (3d Cir.1995), this court held that a proper curative instruction could negate a manifestly improper statement by the prosecutor in the course of closing argument that defendants were guilty of the charged offense. The prosecutor stated: "I suggest you shouldn't believe [the defense witnesses] because they're guilty of exactly the same bankruptcy fraud that these two defendants are guilty of. And don't you assume that they are not going to get what's coming to them either." 47 F.3d at 1264. The court held that the expression of the prosecutor's personal opinion jeopardizes the defendant's right to be tried solely on the basis of the evidence presented to the jury because his opinion "carries with it the imprimatur of the Government and may induce the jury to trust the Government's judgment rather than its own view of the evidence." 47 F.3d at 1265. In this case, like the prosecutor in Zehrbach, the police officer King arguably carried the "imprimatur of the Government."

The court in Zehrbach gave a specific instruction immediately after the objection to disregard the prosecutor's comment, an instruction that the court repeated just a short time later at the close of the prosecutor's argument. The court told the jurors to disregard any personal opinion of counsel and to base their decision solely on the evidence. In its final instructions, the court cautioned the jury that the arguments of counsel were not evidence and that they should not consider any evidence that they were earlier instructed to disregard. This extensive cautioning by the court sufficiently cured the prosecutor's error. 47 F.3d at 1267.

On the last page of the document there's a statement by this witness as to the total value. This witness used the word "fraud." That's his use of the word. This is not the decision of you, ladies and gentlemen. It is your duty and function here to determine whether or not the government has proven beyond a reasonable doubt whether or not this defendant has committed the crimes charged, and the crimes charged have been clearly stated to you. And they are bank fraud and travel fraud. Is this understood? It's his word, not your decision. Understood?

The normal presumption is that a jury will follow the Court's instruction to disregard inadmissible evidence inadvertently submitted to it, "unless there is an overwhelming possibility that the jury will be unable to follow the court's instructions... and a strong likelihood that the effect of the evidence would be devastating to the defendant...." Greer v. Miller, 483 U.S. 756, 766 n. 8, 107 S. Ct. 3102, 97 L. Ed. 2d 618 (1987) (quotations omitted). Thus, in view of these precedents, we believe that the trial judge's strong and unambiguous curative instructions rendered harmless the inadvertent admission of the word "fraud" on the summary of checks cashed, although offered by a government agent.5

Moreover, the admission of the word "fraud," seems trivial in the broader context of the case. See Zehrbach, 47 F.3d at 1267. Officer King's comment is couched within an abundance of evidence of fraudulent conduct. The travel fraud charge requires that the defendant willfully participate in a scheme to defraud someone of property in excess of $5,000 and that the defendant cause or induce someone to travel in interstate commerce in furtherance of the fraud. It is undisputed that Thomas induced Weygandt to sign the checks under false pretenses and that she used the checks for her own purposes, and thus was engaged in a scheme to defraud Weygandt. Even if fraud upon the bank was disputed at trial, thus making King's use of the word "fraud" with respect to the bank fraud charge a debatable issue, there was no real issue as to whether there was a scheme to defraud Weygandt. To prove a travel fraud charge the Government must prove that the defendant induced a person to travel in interstate commerce in furtherance of a scheme to defraud someone of over $5,000. See 18 U.S.C. § 2314. Here, the evidence is strong and clear that Thomas engaged in a scheme to defraud Weygandt of over $5,000 and caused her to cross state lines to visit banks in Delaware as part of that scheme. Indeed, it is difficult to view Thomas's conduct with respect to Weygandt as other than fraud, and the inadvertent error as harmless. See United States v. Copple, 24 F.3d 535, 546 (3d Cir.1994).

We turn now to the appellant's objections to her sentence. Section 3B1.3 of the Sentencing Guidelines provides in part that: "If the defendant abused a position of public or private trust ... in a manner that significantly facilitated the commission or concealment of the offense, increase by 2 levels." U.S.S.G. § 3B1.3. The rule requires a two tier analysis. First the reviewing court must determine whether a position of trust exists. The appellate court reviews the district court's ruling on whether a position of trust exists de novo. The second question, whether that position has been abused, is reviewed for clear error. United States v. Iannone, 184 F.3d 214, 223 (3d Cir.1999).

This Court recently set forth the factors for determining a position of trust: (1) whether the position allows the defendant to commit a difficult-to-detect wrong; (2) the degree of authority which the position vests in the defendant vis-a-vis the object of the wrongful act; and (3) whether there has been reliance on the integrity of the person occupying the position. 184 F.3d at 223.

There is ample evidence to show that Thomas indeed held a position of trust with respect to Weygandt. Thomas argues she was merely a health aide. The evidence, however, shows that the real scope of her job was much broader. There is evidence that Thomas opened Weygandt's mail for her without supervision and that she gave Thomas authority to pay bills for her. These tasks clearly invested Thomas with considerable discretion since Weygandt did not monitor Thomas closely and appeared to rely on her judgment and integrity. The wrong was difficult to detect because Thomas was the person who filled in the amounts and payees on the checks, and Weygandt did not independently verify them. There was substantial reliance on the good faith of Thomas, as there would be in any relationship where financial matters are entrusted to another. These facts satisfy all the elements of this Court's test.

The standard of review for abuse of a position of trust, once it is established that the defendant was in such a position, is clear error. Abuse of trust occurs where the employer or vulnerable party relies on another's integrity for protection against the loss occasioned by the crime, and where the trust aspect of the position made the commission of the crime easier. 184 F.3d at 223-225. There is ample evidence on which to draw such conclusions here. Clearly, Thomas's unique position enabled her to protect Weygandt against financial theft because Weygandt signed checks at Thomas's request and on Thomas's assurances as to their purpose. Moreover, without Weygandt's credulity and trust, it appears likely that the checks would not have been signed, and that Weygandt would not have accompanied Thomas to the bank to vouch for the checks. Thus, to the extent the sentencing enhancement for abuse of trust applies to the travel fraud charge, the District Court committed no error.

Thomas claims that she is subject to a downward sentencing adjustment for acceptance of responsibility, pursuant to U.S.S.G. § 3E1.1. Although she might have had a valid claim that the adjustment applied to her bank fraud conviction, it does not bear at all on her travel fraud sentence. Section 3E1.1 provides that, "If the defendant clearly demonstrates acceptance of responsibility for his offense, decrease the offense level by 2 levels." U.S.S.G. § 3E1.1.

"[T]he District Court's decision whether to grant the adjustment is entitled to `great deference' on review because `[t]he sentencing judge is in a unique position to evaluate a defendant's acceptance of responsibility." United States v. Bennett, 161 F.3d 171, 196 (3d Cir.1998) (quoting U.S.S.G. § 3E1.1 cmt. (n.5)). Here, the District Court, in denying the downward departure, relied chiefly on Thomas's decision to make the Government prove its case at trial. In rebuttal, Thomas points to her pretrial confession to police investigators, which the Government repeatedly referred to in its arguments to the jury, and claims that the Government's factual case was essentially established by virtue of the confession. She further argues that the confession admits to the fundamental wrongdoing and that proceeding to trial was intended not to establish her factual innocence, but to determine whether her conduct fell within the parameters of the bank fraud statute.

The Application Notes do permit a district court to consider truthful admissions to the conduct for which defendant is criminally responsible, so long as no relevant conduct is not falsely or frivolously denied or contested. However, in United States v. DeLeon-Rodriguez, 70 F.3d 764, 767 (3d Cir.1995), this Court held that "a reduction is generally not meant to apply to a defendant who puts the government to its burden of proof at trial." The Application Notes from the Guidelines add that only "[i]n rare situations a defendant may clearly demonstrate an acceptance of responsibility for his criminal conduct even though he exercises his constitutional right to a trial. This may occur, for example, where a defendant goes to trial to assert and preserve issues that do not relate to factual guilt (e.g., to make a constitutional challenge to a statute or a challenge to the applicability of a statute to his conduct)." U.S.S.G. § 3E1.1 cmt. (n.2).

Although there may have been a good faith challenge to the applicability of the bank fraud statute to Thomas's conduct, there could have been no serious doubt as to the applicability of the travel fraud statute. The Government charged that Thomas had engaged in a scheme to defraud Weygandt and, in furtherance of that scheme, had taken Weygandt across state lines. These allegations were indisputably covered by the travel fraud statute. Thus, while there was no colorable legal defense to the travel fraud charge, Thomas nonetheless forced the Government to prove its case at trial. Thus, this is not the "rare situation" where a defendant did accept guilt, despite seeking a trial. See U.S.S.G. § 3E1.1 cmt.(n.2). The District Court committed no error in rejecting the appellant's challenge.

In summary, we hold that the relevant requirements under the bank fraud statute are: a defendant must execute, or attempt to execute, a scheme or artifice, intended to victimize a federal bank or federally insured bank by causing it an actual or potential loss of its own funds. Where the scheme involves the mere withdrawal of funds in the bank's custody, the Government must show that the withdrawal exposed the bank to some form of liability as a result of the fraud. There was none here.

Accordingly, because there is no proof that Thomas intended to victimize the banks or that the banks suffered a loss, the Government's case as to bank fraud fails as a matter of law on Count I. The admission of the document containing the word "fraud" is harmless error, and we will affirm the District Court's judgment as to Count II. We also hold that the District Court correctly determined the sentencing issues before it, to the extent those issues pertained to the travel fraud count. The judgment of conviction and sentence will be reversed on Count I and the case remanded to the District Court for further proceedings and resentencing consistent with this opinion.