Restaurant Data Breach Probe Filing: Card Data In Plain Text, Default Passwords And Wide Open Wireless Access

A Massachusetts restaurant chain, which was just fined $110,000 by that state's attorney general as a result of a substantial data breach, is a textbook example of how not to handle payment security. Court filings from the case paint a classic picture: unchanged default passwords, wide open wireless access, full card data stored in plain text and an impressive lack of concern about the breach, with restaurants continuing to accept payment cards after the chain knew of the breach and malware that had not yet been deactivated.

The breach at the chain, The Briar Group (The Lenox, MJ O'Connor's, Ned Devine's, The Green Briar and The Harp), impacted at least 125,000 MasterCard and Visa customers, the state filing said. Other security naughtiness alleged: using the default usernames and passwords from its Micros POS systems, opting to not change network passwords "for more than five years," allowing those username/password combos to be "used system-wide for all users" and then not changing passwords after employees quit or were fired.

Briar also "granted all of its employees local administrator access on their assigned desktop computers, thus imposing no limitation on the software that could be installed on corporate equipment," the state said in a civil complaint.

The situation apparently began on April 24, 2009, when cyberthieves installed code on a Briar's server to capture full Track 1 and Track 2 data, according to excerpts from a report by Verizon Business Network Services, which was retained to conduct a standard forensic investigation after the chain learned of the breach.

The incident was apparently discovered on Oct. 15, 2009, when an unidentified European processor detected payment card fraud and the common points of purchase pointed back to The Briar Group. On October 20, Visa contacted Briar's acquirer—Sovereign Bank—which in turn reached out to First Data, which managed transactions for Briar. Briar received official notice of the breach from First Data on Oct. 29, 2009. A couple of weeks later—on November 13—Citigroup notified Briar of more fraudulent activity.

The Verizon probe's initial report suggested that the malware's creator was rather brazen, in that there seemed to be little attempt to hide what was being done. For example, one file in plain view on the server was labeled, according to the state filing, "C:\\WINDOWS\system32\memdump on each of Briar's six compromised Micros 9700 Point of Sale servers." Added the report: The program "parsed through the memory dumps [looking] for payment card data and saved it in an output file '' that contained Tracks 1 and 2 data."

The code was finally removed from Briar's systems on Dec. 10, 2009, the filing said.

The state complaint detailed how it thinks the program spread: "The intruder was able to access Briar's home server and from there was able to access the other restaurant locations and install malcode, because all of the affected locations were constructed on a flat network on a single Windows domain. A flat network is a network in which workstations are directly connected to one another, without intermediary hardware devices. The Micros 9700 POS servers had remote access utilities, including pcAnywhere and Microsoft Remote Desktop, enabled at the system start. The open configuration of these utilities allowed access of the Micros 9700 servers from the Internet."

Massachusetts, citing the Verizon report, also didn't care for Briar's wireless network setup.

"Wireless networks were available at each of the affected locations, but did not have any security enabled on them, such as Wi-Fi Protected Access," the complaint said. "The wireless and wired networks connected to a firewall, but were not properly segmented, as the Micros POS system was reachable once a user connected to Briar's wireless network."

Although the unencrypted payment card data is a horrible security practice, the state filing said it couldn't prove that any unencrypted data made its way to the cyberthieves. Verizon "discovered 7,130 instances where unencrypted cardholder account numbers and expiration dates, in clear text, were retained," the filing said.The state also pointed out that Briar was not PCI compliant at the time of the incident, but details—such as whether Briar attempted to be declared PCI compliant—were unclear. Even after changing its password-management approach, "Briar failed to validate its PCI compliance until April 21, 2010, almost six months after it learned of the data breach, nearly a year after malcode was installed, and only after it was fined by Visa and MasterCard for continued non-compliance with PCI DSS." Presumably, the state meant a fine was accessed to the processor, which politely passed it along to the retailer.

The state's filings are consistent on the laundry list of security violations that make TJX's breach resemble Fort Knox. But elsewhere in the filings, inconsistencies crop up that raise fundamental questions about the legitimacy of the state's charges.

For example, the state quoted from a Nov. 5, 2009, E-mail from Briars President Austin M. O'Connor. In that E-mail, "Briar's president noted that he wanted 'to do the right thing' but did not want to have to 'pay for an investigation that [Briar] could somehow avoid.'" The state's characterization was that the executive "initially expressed reluctance to hire a computer forensics company."

A few concerns: In the complaint, it quotes O'Connor as saying "pay for an investigation that they could somehow avoid." In context, the "they" has to refer to Briar. But in what possible context could the Briar president have referred to his company as "they"? Whether it was written to a fellow Briar employee or an outside contractor, it's hard to envision. The state didn't say who the E-mail was written to or provide any other context.

Of far greater concern with that E-mail quote is the point the state is trying to make. Clearly, all executives don't want to pay for probes (or, for that matter, pay taxes or take out the garbage at home). That doesn't mean they won't do it, nor does it mean they hesitate to do it. If the state's point was that O'Connor delayed the probe or somehow weakened the chain's defense against the breach, it would have said so. The fact that it didn't makes it unclear what it's alleging with the quote.

The implication is that O'Connor caused a delay in the probe, but the timeline in the filing doesn't support the contention of a delay and it isn't even consistent. For example, one state filing said that "three weeks after Briar was notified by First Data of the breach, and after being required by Visa to retain a Qualified Incident Response Assessor, Briar engaged Verizon" and the next sentence said that Verizon "arrived at Briar's headquarters on or around Nov. 15, 2009, to begin its investigation."

Given that Briar received its first notice (from First Data) on October 29, three weeks later would place it on or about November 19. And yet the complaint has Verizon arriving at Briar to begin the investigation on or around November 15. Given that Verizon would have presumably arrived after it had been engaged, this is very confusing.

The state filing also references—without explanation—that "more than 125,000 consumers were harmed by Briar's conduct." What was the nature of that harm? There's a reference to a similar number earlier ("over 53,000 MasterCard and over 72,000 Visa accounts were affected by the data breach, providing a channel for the infliction of fraud on consumers and forcing consumers and their banks to endure the inconvenience, cost and uncertainty of cancelling and re-issuing thousands of credit and debit cards used at Briar."

But that first number is merely saying how many numbers were involved and spoke of the possibility that some might be impacted ("providing a channel for"). Even if consumers were impacted, wouldn't they be fully compensated by zero-liability programs from both MasterCard and Visa? The "cancelling and re-issuing" is a bank—as opposed to a consumer—impact. State officials defined what it meant by "consumer harm" as a situation where consumers' data is put at risk.