A while ago, I wrote a post about the US government’s watchlisting system and how it prompted Hina Shamsi, director of the American Civil Liberties Union’s National Security Project to conclude that “we’re getting into Minority Report territory”.
This week, I have an even more frightening proposition. Remember Skynet, the “fictional, self-aware artificial intelligence system” from Terminator that “serves as the franchise’s main antagonist”? That’s the territory we may be getting into.
No, I’m not joking. Edward Snowden in his latest interview with James Bamford for Wired spoke about a previously unreported programme called MonsterMind which Snowden says he “discovered while getting up to speed on the capabilities of the NSA’s enormous and highly secret data storage facility in Bluffdale, Utah.”
MonsterMind, according to Wired was “the last straw for Snowden” and, its frightening potential aside, it paints a telling picture regarding the US’ ambition of dominating the internet.
No humans involved: Inside the Monster’s Mind
So first of all, what is MonsterMind and what can it do?
Described as “the NSA bot that could wage Cyberwar autonomously,” Monstermind
would automate the process of hunting for the beginnings of a foreign cyberattack. Software would constantly be on the lookout for traffic patterns indicating known or suspected attacks. When it detected an attack, MonsterMind would automatically block it from entering the country—a “kill” in cyber terminology.
Kim Zetter suggests that we
Think of it as a digital version of the Star Wars initiative President Reagan proposed in the 1980s.
Similar to Star Wars shooting down incoming nuclear missiles, Monstermind would “shoot down” an incoming cyber-attack before it entered the country (the country being the US of course).
Fair enough, you might think. Surely, there is nothing wrong with trying to protect your country against Darth Vader attacking you on the virtual plane and besides, “[p]rograms like this ha[ve] existed for decades.”
That may be so. However, MonsterMind wouldn’t be just another programme. Rather, its software
would add a unique new capability: Instead of simply detecting and killing the malware at the point of entry, MonsterMind would automatically fire back, with no human involvement.
So, if this was Reagan’s Star Wars, after shooting down the incoming missile, the programme would then shoot back at the missile’s point of origin automatically – with no human middle-man to double check that shooting back is the wisest idea. You might have guessed what the problem is here but let’s spell it out just the same:
The initial attacks are often routed through computers in innocent third countries.
So, if this was the Star Wars initiative, the scenario would play out somewhat like this: A computer detects an incoming missile. The computer automatically shoots the missile down. It then launches its own missile without anyone double checking where exactly the first missile originated from. Problem: the incoming missile was not fired from any facility directly connected to whoever was behind the attack but from a place that the attacker had temporarily highjacked to stop themselves being detected. So any automatic retaliation by your friendly programme wouldn’t actually target the attacker but the proxy behind which they were hiding. That proxy could, for example, be a critical civilian structure, like a hospital. Or in fact a heavily armed military facility of a hitherto non-hostile party.
For the virtual world, this is how Kim Zetter describes it:
An attack from a foreign adversary – (i.e. the equivalent of the initial missile strike) – likely would be routed through proxies belonging to innocent parties—a botnet of randomly hacked machines, for example, or machines owned by another government.
Or, as Edward Snowden told James Bamford:
You could have someone sitting in China, for example, making it appear that one of these attacks is originating in Russia. And then we end up shooting back at a Russian hospital. What happens next?
Let’s imagine it: “unanticipated collateral damage” or retaliation from a previously non-hostile third party. Boom, nuclear war, everyone dies. If you are noticing certain parallels to aforementioned works of fiction, feel free to indulge them.
Granted, as far as MonsterMind is concerned the complete annihilation of humanity is not the default consequence. As Kim Zetter points out,
Snowden doesn’t specify the nature of the counterstrike to say whether it might involve launching malicious code to disable the attacking system, or simply disable any malicious tools on the system to render them useless.
depending on how its [sic] deployed, such a program presents several concerns, two of which Snowden specifically addresses in the WIRED story.
Privacy under attack – again
Bamford describes it thus:
…in order for the system to work, the NSA first would have to secretly get access to virtually all private communications coming in from overseas to people in the US.
All private communications. That is everything, from everyone, everywhere, all the time. Not just suspects.
Why is that? Well. “The argument,” Snowden tells Bamford, “is that the only way we can identify…malicious traffic flows and respond to them is if we’re analyzing all traffic flows.”
So, in order to respond to an attack, the system first needs to identify the attack. To be able to do that, so the argument goes, the system needs to be able sift out a potential attack from everything else that is going on. For this is has to “seiz[e] private communications without a warrant, without probable cause or even a suspicion of wrongdoing.” And back we are in violating-the-fourth-amendment-territory again.
So, here is a programme that not only, allegedly, needs to monitor everybody’s communications all the time to be able to work but which also, potentially, could fire back on its own, thus damaging critical (civilian) infrastructure or starting an outright war. Ain’t it shiny?
Mapping Cyberspace: Plan X and Bonesaw
It gets scarier.
Kim Zetter notes the resemblance between MonsterMind and effort called Plan X, reported by the Washington Post in 2012:
Plan X is a project of the Defense Advanced Research Projects Agency, a Pentagon division that focuses on experimental efforts.” As one of these efforts, the goal with plan Plan X was to “dominate the digital battlefield just like…the traditional battlefield.” As part of this, the US sought (or is seeking) to create “an advanced map that details the entirety of cyberspace — a global domain that includes tens of billions of computers and other devices — and updates itself continuously.
Let that sink in for a moment: a map of all of cyberspace. That’s pretty ambitious and more than a little colonialist.
One such map already seems to exist. James Bamford in a 2013 article for Wired mentions
Bonesaw, [a] map [that] displays the geolocation and digital address of basically every device connected to the Internet around the world, providing what’s called network situational awareness.
Among other things it displays
what software is running on the computers inside [a given target facility], what types of malware some may contain, and a menu of custom-designed exploits that can be used to secretly gain entry. It can also pinpoint those devices infected with malware… as well as networks turned into botnets and zombies— the equivalent of a back door left open.
Basically, imagine a map that shows a burglar your neighbourhood while at the same telling the burglar that you’ve left the backdoor unlocked.
There’s more to mapping than just efficient burgling though. In post-colonial studies, “mapping” is seen as “an act of mastery and control” that contributes to the “establishment and re-establishment” of a certain world orders. Maps were not only “used to assist in the process of aggression,” but also to claim ownership over certain territories and establish a set of dominant rules, ultimately allowing the establishment of a norm, deviance from which was constructed as threatening and other. And we all know what happens to what is perceived as the threatening Other: it must be contained, by force if need be.
The parallels between colonial mapping and the effort of mapping – and thus dominating – the internet as a potentially democratic space, even a ‘diasporic utopia’ of innovation, dissent, resistance and anarchy, in which the status quo may be challenged and subverted, are immediately obvious. Governments the world over are constantly striving to dominate at least their corners of the internet and to curb its potential for free speech and political dissent. A similar desire for control is evident in proposals to give “the United Nations the power to organise and supervise the internet or to grant such authority to the International Telecommunications Union.”
Mapping cyberspace, it appears, would go a lot further in enforcing the hegemony of the mapper than mapping physical space would: it would allow those in possession of the map to identify malicious entities, locate them and act against them (surveillance delivers the knowledge at the basis of physical acts of warfare). The problem with that of course is that those in possession of map and control might have a different view from anyone else about what is considered malicious. Far from limited to the much-evoked terrorism, this could simply be peaceful disagreement with the dominant ideology.
Currently, “the California-based Internet Corporation for Assigned Names and Numbers (Icann)…is subject to US law, is contracted by the US administration and is empowered to supervise how digital traffic operates.” The US-centric nature of current internet governance has been criticised repeatedly in the wake of the Snowden revelations.
It is little surprising that the US is unwilling to give up what dominance it has over the internet, or that it attempts to increase or secure that dominance in secret.
Science without the Fiction: Plan X, MonsterMind, Skynet
For governments or their institutions – most notably the intelligence agencies and militaries – the internet is both threat and opportunity and efforts like MonsterMind and Plan X aren’t just about defence or even dominance, but about attack: Plan X architects
hope[d] to develop systems that could give commanders the ability to carry out speed-of-light attacks and counterattacks using preplanned scenarios that [did] not involve human operators manually typing in code — a process considered much too slow…Officials compare this to flying an airplane on autopilot along predetermined routes.
I cannot be the only one to whom this sounds like the virtual equivalent of a drone. Tellingly, Plan X is a project of the Defense Advanced Research Projects Agency, whose works also “gave rise to stealth jet technology and portable global-positioning devices.”
Also, get this:
Skynet was a computer system developed for the U.S. military…as a “Global Digital Defense Network” and given command over all computerized military hardware and systems, including the B-2 stealth bomber fleet and America’s entire nuclear weapons arsenal. The strategy behind Skynet’s creation was to remove the possibility of human error and slow reaction time to guarantee a fast, efficient response to enemy attack.
My emphasis – compare the “no more human operators manually typing in code” under Plan X, and MonsterMind as a programme that might automatically launch counter attacks.
Well, we all know what happened with Skynet, don’t we? It started operating on its own, came to the conclusion that not only did it not need humanity but that humanity was actually a threat to it and launched – wait for it – “nuclear missiles under its command at Russia, which responded with a nuclear counter-attack.”
Great. Yes, Terminator is fictional. But once we switch the DVD off, we return to a real word in which “the spy agencies are helping drive a lucrative, dangerous, and unregulated cyber arms race, one that has developed its own gray and black markets”, a world in which the dystopian of Orwell and Kafka don’t seem so far away anymore.
So much for the “fiction” side of science fiction.