In a long-awaited decision (at least by the parties and fidelity law practitioners) the Fifth Circuit Court of Appeals has held that a “Computer Fraud” Insuring Agreement in a Crime Insurance Policy does not cover the insured’s loss after its employees were tricked into wiring approximately $7 million to a fraudulent bank account set up by criminals posing as one of the insured’s trusted vendors. Apache Corp. v. Great American Insurance Co., No. 15-20499, 2016 WL 6090901 (5th Cir. Oct. 18, 2016).

This case arises from a 2013 social engineering scheme1 whereby several criminals, suspected to be based in Latvia, posed as representatives of Petrofac, a vendor of Apache, the insured oil-production company. After corresponding with the fraudsters by phone and over email, the imposters convinced Apache to direct payments to a “new” bank account for Petrofac. In brief, the scheme unfolded as follows:

  • An Apache employee received a telephone call from a person identifying herself as a representative of Petrofac, Apache’s vendor. The caller instructed Apache to change the bank account information for its payments to Petrofac. The Apache employee responded that such a change could not be accomplished without a formal request on the vendor’s letterhead.
  • A week later, Apache’s accounts payable department received an email from a “” address2 in which the sender advised that the purported “new” bank account was effective and payments going forward should be made to the new account. The sender attached a letter to the email with both the old and “new” bank account information. This letter was on Petrofac letterhead and contained a signature by a purported Petrofac employee.
  • In response to the email with attached letter, an Apache employee called the number listed on the letter. An individual posing as a Petrofac employee purportedly confirmed the request to change the banking information.
  • An Apache employee then entered an internal “change request” with the new Petrofac bank account information. A separate Apache manager then approved the request and thereafter payments began flowing to the new bank account, which was really a fraudulent account set up by the imposters.

The scheme unraveled when the “real” Petrofac contacted Apache regarding several delinquent invoices. Upon investigation, Apache realized that it had been duped. Fortunately, Apache was able to recover a substantial portion of the funds, leaving its total loss at only $2.4 million, of which Apache claimed a loss of approximately $1.4 (after application of the deductible) under the “Computer Fraud” provision of its crime policy.

The Computer Fraud Insuring Agreement

Apache’s crime policy contained, in part, the following “Computer Fraud” Insuring Agreement:

We will pay for loss of, and loss from damage to, money, securities and other property resulting directly from the use of any computer to fraudulently cause a transfer of that property from inside the premises or banking premises: (a) to a person (other than a messenger) outside those premises; or (b) to a place outside those premises.

After analyzing Apache’s loss, Great American denied coverage, in part, because the loss “did not result directly from the use of a computer nor did the use of a computer cause the transfer of funds.” Essentially, the insurer examined the fraud and found that the email did not, by itself, cause the funds transfer and coverage under the Computer Fraud agreement was limited to hacking and unauthorized computer use. Apache disagreed and sued the insurer, claiming that the Computer Fraud agreement said nothing about “hacking” and the denial was improper because the email from the fraudsters constituted computer fraud which directly caused the fraudulent transfer of funds.

The District Court Decision

In August 2015, the U.S. District Court for the Southern District of Texas handed down a decision granting Apache’s motion for summary judgment, noting that “the intervening steps of the [post-email] confirmation phone call and supervisory approval do not rise to the level of negating the email as being a substantial factor” in bringing about the loss, thus presenting a factual scenario sufficient to allow coverage under the “directly resulting from” language in the Computer Fraud Insuring Agreement. According to the district court, “despite the human involvement that followed the fraud, the loss still resulted directly from computer fraud, i.e., the email directing Apache to disburse payments to a fraudulent account.”

The Fifth Circuit Reverses and Gets it Right

On October 18, 2016, the Fifth Circuit vacated the district court decision and entered judgment in favor of Great American, noting that “there is cross-jurisdictional uniformity in declining to extend coverage when the fraudulent transfer was the result of other events and not directly by the computer use.” After surveying a spate of recent case law involving similar disputes, the court found that the now vacated district court decision was the “only presented decision interpreting the computer-fraud policy language to cover a loss when the computer use at issue was limited to email correspondence.” The Court found that the email Apache highlighted in order to assert coverage under the Computer Fraud agreement was undoubtedly “part of the scheme,” but merely incidental to the occurrence of “the authorized transfer of money.”

Note that the email, by itself, caused no transfer of money; rather, it was just one step in a multi-step process that ended in Apache failing to detect the imposters and making large payments to a fraudulent bank account. To interpret the computer fraud provision as reaching any fraudulent scheme involving an email communication would, as stated by the court, “convert the computer-fraud provision to one for general fraud,” a risk the insurer clearly did not contemplate.3 Essentially, the Court read the “resulting directly from” language in the Computer Fraud agreement to require the identified “use of a computer” to cause the fraudulent transfer itself (i.e., a true hacking incident), not merely set in motion, or be part of, a chain of events whereby an insured fails to investigate the accuracy of fraudulent information provided to it.


Social engineering schemes similar to that described above are on the rise because of the ease in which fraudsters can obtain information about a target online. By using a combination of telephonic and online communication, even large companies are vulnerable to sophisticated schemes of fraud. In response to these threats, certain carriers have introduced specialized extensions of coverage or separate products aimed at covering “social engineering” risks. Nonetheless, certain coverage directed at true “hacking” threats, as was the intent of Great American’s insuring clause here, will likely be subject to challenges from insured’s lacking appropriate “social engineering” coverage. The Fifth Circuit’s Apache decision provides helpful authority to define the limits of standard computer fraud wording as it appears in crime policies and shows that the use of a computer somewhere in the chain of a multistep fraudulent scheme is insufficient to trigger coverage.