The Atlantic Monthly just published a piece about the computer security challenges facing universities. Those challenges are serious:
“Universities are extremely attractive targets,” explained Richard Bejtlich, the Chief Security Strategist at FireEye, which acquired Mandiant, the firm that investigated the hacking incident at the [New York] Times. “The sort of information they have can be very valuable — including very rich personal information that criminal groups want, and R&D data related to either basic science or grant-related research of great interest to nation state groups. Then, on the infrastructure side they also provide some of the best platforms for attacking other parties—high bandwidth, great servers, some of the best computing infrastructure in the world and a corresponding lack of interest in security.”
The issue is framed in terms of “corporate lockdown” vs. “bring your own device,” with an emphasis on network security:
There are two schools of thought on computer security at institutions of higher education. One thought is that universities are lagging behind companies in their security efforts and need to embrace a more locked-down, corporate approach to security. The other thought holds that companies are, in fact, coming around to the academic institutions’ perspective on security—with employees bringing their own devices to work, and an increasing emphasis on monitoring network activity rather than enforcing security by trying to keep out the outside world.
There’s a nod to application security, and it’s actually a great example of setting policies to incentivize users to set stronger passwords (this is not easy!):
A company, for instance, may mandate a new security system or patch for everyone on its network, but a crucial element of implementing security measures in an academic setting is often providing users with options that will meet their needs, rather than forcing them to acquiesce to changes. For example, Parks said that at the University of Idaho, users are given the choice to set passwords that are at least 15 characters in length and, if they do so, their passwords last 400 days before expiring, whereas shorter passwords must be changed every 90 days (more than 70 percent of users have chosen to create passwords that are at least 15 characters, he added).
Getting hacked is about losing control of one’s data, and the worries in the last two passages have to do with things the university can’t directly control: the security of devices that users connect to the network, and the strength of passwords chosen by users. Things beyond one’s control are generally anxiety-provoking.
Taking a step back, the data that’s of interest to hackers is found in databases, which are frequently queried by web applications. Those web applications might have vulnerabilities, and a university’s application code is under the university’s control. The application code matters in practice. Over the summer, Team GhostShell dumped data stolen from a large number of universities. According to Symantec:
In keeping with its previous modus operandi, it is likely that the group compromised the databases by way of SQL injection attacks and poorly configured PHP scripts; however, this has not been confirmed. Previous data dumps from the 2012 hacks revealed that the team used SQLmap, a popular SQL injection tool used by hackers.
The attack, which has implications for the integrity of USyd’s security infrastructure, compromised the personal details of approximately 5,000 students and did not come to the University’s attention until February 6.
The hacker, who goes by the online alias Abdilo, told Honi that the attack had yielded email addresses and ‘pass combo lists’, though he has no intention of using the information for malicious ends.
“99% of my targets are just shit i decide to mess with because of southpark or other tv shows,” he wrote.
As for Sydney’s breach on February 2, Abdilo claimed that he had very little trouble in accessing the information, rating the university’s database security with a “0” out of 10.
“I was taunting them for awhile, they finally figured it out,” he said.
That’s a nightmare, but solving SQL injection is a lot more straightforward than working on machine learning algorithms to improve intrusion detection systems. Less academic, if you will. The challenges in the Atlantic article are real, but effective measures aren’t always the most exciting. Fixing legacy applications that students use to look at their grades or schedule doctor’s appointments doesn’t have the drama of finding currently-invisible intruders. There is no incentive or process for the faculty to work on improving the software development life cycle when their priority is often research productivity or serving on administrative committees. In these instances, partnering with a third-party SaaS application security provider is the best solution for these urgent needs.
In a sense, it’s good news that some of the most urgently needed fixes are already within our abilities.