NSA Director Says Cyberattacks on Critical Systems a Matter of “When, Not If”

I recently had the opportunity to review the entire testimony of Adm. Michael Rogers, director of the National Security Agency (NSA) and head of U.S. Cyber Command, to the House Intelligence Committee hearing, available at C-SPAN. It seems the purpose of the testimony was to support an information-sharing bill. Now, I prefer to focus on intrusion prevention rather than sharing information about already-detected intrusions, but still I found that the admiral said a number of interesting things relevant to modern intrusions and the capabilities of our adversaries.

For example, Admiral Rogers said, among other things, in response to a question about the capabilities of “trojan horses” found on industrial control system (ICS) networks:

“There shouldn’t be any doubt in our minds that there are nation-states and groups out there that have the capability to do that – to enter our systems, to enter those industrial control systems, and to shut down – forestall our ability to operate – our basic infrastructure.”

This statement was big news, given that the admiral is the highest-ranked individual in the American administration to have admitted that our critical infrastructure could be hacked. But to people working in the critical infrastructure cybersecurity field, this is not news at all. Common wisdom has it that any site can be hacked if an adversary is given enough time, enough money and enough talent to do the hacking – and nation-states generally have all three in abundance.

The growing threat of nation-states

What was more interesting to me was when Admiral Rogers elaborated on this statement. The headlines that followed his testimony were all about China having the ability to shut down critical infrastructures, but the Admiral’s comments were clear – several nation-states have this capability and others are developing it, and other groups and even individuals are doing the same. For example, Admiral Rogers said that his agencies are seeing criminal gangs starting to use the tools and techniques that have historically been attributed only to nation-states. It would appear that some nation-states are outsourcing their cyberattacks. Organized crime has a long history in the cyber-security world and is responsible for the majority of malware and botnets which plague all computers connected to the Internet. The question we’d all like to see answered is, “what else will these criminal groups use these types of attack techniques for, and when?”

Admiral Rogers repeatedly gave the example of the Shamoon malware, which erased 30,000 computers on the Saudi Aramco corporate network. Erasing hard disks on a control system network is a comparatively low-tech attack, but it is unfortunately very likely to be an extremely effective attack. Modern infrastructure generally cannot be operated without human oversight, and control system computers are essential in providing such oversight. Erase enough control system hard drives and the physical critical infrastructure – the power plant or pipeline – must be shut down.

How long will it take to bring back up? The Admiral was vague here, and for good reason. How long a site takes to recover from a Shamoon-style attack on control system computers very much depends on the physical industrial process in question, and the recovery time depends on how thorough and how well-practiced our disaster recovery plans are. Do we have current back-ups for every part of the control system? Were any programmable logic controllers (PLCs) or other devices attacked and erased? Do we have back-ups of that equipment?

Information sharing alone is insufficient

Now, the focus of the Admiral’s testimony was the current information-sharing bill, and so “information sharing” was the remediation that he returned to time and again when questioned. I believe that information sharing is a good thing, but it is far from sufficient in terms of preventing a widespread outage of critical infrastructures. Information sharing only works after we have discovered the characteristics of a compromise so surviving infrastructure sites can try to detect similar compromises before they, too, are crippled.

Information sharing does little to prevent widespread, simultaneous compromise. Imagine, for example, a bit of malware disguised as a device driver security update-checking program. The program looks harmless – it reaches out to a plausible-seeming website periodically to check for updates. (For the record, there should be no route from control computers to the Internet to begin with, but that rule of security gets broken more often than not.) Of course, the website is a sham, and when this bit of malware downloads and runs a particular update, suddenly hundreds or even thousands of infrastructure sites malfunction simultaneously. Did information sharing save us?

There is obviously a time and a place for information sharing, but for most critical infrastructure ICS networks, strong intrusion prevention is more important than information sharing. Furthermore, since it is theoretically impossible to reliably ask some firewall or other intrusion detection software to differentiate “good software” from “bad software” (or even “good messages” from “bad messages”), hardware-enforced Unidirectional Security Gateways at critical infrastructure cyberperimeters are one of the few very effective tools we have at our disposal to defeat these modern threats and persistent, remote attack patterns.

Applying new cybersecurity best practices

Strong cyberperimeter protections must be part of the security response to these critical ICS threats. Unidirectional Security Gateways are the new industrial cybersecurity best practice, most recently included in the new ANSSI cybersecurity guidelines. Information sharing is a worthwhile program, but it will not save us if all we have protecting our critical networks is software.

Want to hear the latest news impacting industrial security? Follow Waterfall Security on Twitter: @WaterfallSecure.