The Indian Air Force chief on Monday said that India’s defence networks, sealed off from the internet, may face danger from insiders, a weakness that has undone some of the best-guarded computers in the world.
“Our networks are air-gapped, so people cannot do cyber attack on that. So, insider threat is being generated. We are aware of that,” Air Chief Marshal Ashutosh Dixit said at the ANI National Security Summit 3.0 in Delhi.
He said the defence industry was alert to cyber espionage and insider threats, and that “some technical solutions and some institutional solutions are already in place” across public sector units, laboratories of the Defence Research and Development Organisation (DRDO), the government’s military research agency, and some private firms.
Air gaps
An air-gapped computer, or network of computers, has no connection of any kind to the internet or to any outside network. The only way to get information onto it, or off it, is for someone to take it across by hand, on a disc or a USB stick. A hacker anywhere else in the world has no direct connection to come in through.
Militaries use the arrangement for their most sensitive systems, so do stock exchanges, insurers and biomedical manufacturers, according to Israeli researchers who study how such networks are breached.
Dixit did not say how India’s networks are sealed off. But cutting the cable has never been enough on its own to keep a secret, as the US found out during the second world war, long before there was an internet to cut.
Secret rooms
During the second world war, the US military sent its most sensitive messages through cipher machines, electrical devices that scrambled each letter into code, which could only be turned back into readable text by someone who held the right key at the other end.
In 1943, a researcher at Bell Labs noticed that every time one of these machines scrambled a letter, an oscilloscope in a distant corner of the lab jumped. An oscilloscope draws electrical signals as a moving line on a screen, and this one was picking up the cipher machine itself. Whenever the machine’s parts switched to scramble a letter, they gave off a small burst of radio energy, and each burst matched a letter going in. Anyone who studied the pattern could read the message before it had been scrambled at all.
Bell’s engineers then listened in on a working army cipher centre from across the street, about 80 feet away. Within a few hours, they had recovered about three-quarters of its messages.
The US gave this kind of leak a codename, TEMPEST, and more than 80 years on, its intelligence agencies still build their most secret rooms to stop it. These rooms, known as sensitive compartmented information facilities, or SCIFs, have doors that must meet TEMPEST standards and walls soundproofed well enough that a conversation inside cannot be overheard from the corridor.
A phone is kept inside only with an official’s approval, and a smartphone counts as high risk. In some areas, even copying files onto an approved disc needs permission in advance.
The gap has been crossed before
In 2010, security researchers found a computer worm, a program that copies itself from one machine to the next, that travelled on USB drives. It came to be called Stuxnet, and it remains the most famous breach of an air gap.
Symantec, the US security firm that took it apart, found that it exploited a flaw that let it run on its own from an infected drive, without anyone launching it. Most of the infected computers were in Iran, and the worm was built to attack one particular industrial control system, the kind of computer that runs machinery in plants and pipelines.
Symantec could only guess how it reached that system, and its guess was a removable drive brought in by “a willing or unknowing third party”. The New York Times reported in 2012, citing officials, that the US and Israel had built Stuxnet together under a programme code-named Olympic Games, aimed at Iran’s nuclear programme.
Two years before Stuxnet surfaced, the US military had been on the receiving end.
At a base in West Asia, someone pushed a flash drive, the small USB stick people use to move files between computers, into a military laptop. The drive held malicious code “placed there by a foreign intelligence agency”, William Lynn, then the US deputy secretary of defence, wrote later.
The code uploaded itself onto a network run by US Central Command, which oversees American forces in West Asia. From there it spread undetected through classified and unclassified systems alike, opening a route for data to flow out to servers under foreign control.
Lynn called it “the most significant breach of US military computers ever”. He named neither the agency nor whoever pushed the drive in.
What an adversary could not achieve remotely, jumping the air gap through a hack, was in the end carried out by someone who walked the code across that gap in the real world.
Mr Robot, the hacker drama, built a whole scene around that weak point in its first season in 2015. Darlene, one of its hackers, wants into a prison’s computer network, so she scatters USB sticks loaded with malicious code across the prison car park and waits for an officer reporting for work to pick one up and plug it in.
Researchers at the University of Illinois Urbana-Champaign tested the trick on their own campus, leaving 297 flash drives lying around, and people plugged in 48% of them. The first went in less than six minutes after it was dropped, and most of those who plugged one in had been hoping to find its owner.
The show’s hackers bring their own hardware too. Elliot, its lead hacker, gets past security at Steel Mountain, a fictional high-security data centre, and hides a Raspberry Pi, a credit card-sized computer, on its climate-control system, so the group can later turn up the heat and destroy the backup tapes inside.
At Nasa, a real Raspberry Pi did actual damage. Someone had connected one to the network of the agency’s Jet Propulsion Laboratory without authorisation, and in April 2018 a hacker used it to get in, Nasa’s inspector general found. The intruder stayed hidden for about 10 months and got into two of the laboratory’s three main networks. In all, about 500 megabytes of data were taken.
JPL’s network was not air-gapped, though the lesson applies to sealed networks all the same, since a gadget plugged in without approval can open a door as surely as a stray USB stick.
Researchers have also shown that malicious code inside a sealed network can also send data back out without any cable. They have leaked data through sound too high for people to hear, through radio emissions, through heat passed between computers standing side by side and through a blinking light watched by a camera.
But these methods are slow. One radio technique sends up to 60 bytes a second, about a line of text. Every one of them also needs malicious code already running inside the network, and someone has to have brought it in first.
The ‘insider’
In every case above, the attacker still needed a person, knowing or unknowing, to bridge the gap, whether by bringing in a drive or by plugging in a device that had never been approved.
Once there is no wire for an adversary to come down, the cheapest way into a sealed system is through somebody who already has access to it. That is the threat Dixit described on Saturday.
“We are keeping a close eye; there is a systemic effort to control this,” he said.
































