In a horrifying twist, the NSA may still be working on an artificial intelligence codenamed MonsterMind. Reminds me of the beast/666 computer rumors.
In a recent Wired interview Edward Snowden revealed among other things that the NSA was (and may still be) working on a project codenamed MonsterMind. The autonomous AI would monitor all data traffic entering the U.S. from anywhere in the world looking for potential cyber-attacks. It would then shut down the attack and could launch automated counter-attacks.
Of course, this would require analyzing all network traffic in order to design an algorithm that distinguishes normal traffic flow from malicious traffic.
And as Snowden told WIRED’s James Bamford, "That means violating the Fourth Amendment, seizing private communications without a warrant, without probable cause or even a suspicion of wrongdoing. For everyone, all the time."
(Score: 5, Interesting) by martyb on Monday August 18 2014, @02:38AM
Okay, so the NSA creates this system to "fight" perceived threats. I'd wager that with this being publicized that other actors, if they have not already, are considering implementing something similar. For the sake of argument, imagine that Russia and China have implemented something similar.
From TFS:
Let's try a hypothetical situation.
The NSA's MonsterMind\ detects an attack from, let's say China, and automatically acts to shut it down and launch an automated counter-attack. One would expect the response to not come directly from NSA headquarters. Rather, it would be from some other locations scattered about the internet. Further, in an effort to provide some plausible deniability, I wouldn't be surprised if some amount of the response is routed from/through other nation's infrastructure. Fire up some bots in Russia and have those attack targets in China.
Meanwhile, the Chinese system detects these counter-attacks and reacts in the same way; it tries to shut down the attacks and launch counter-attacks. The Chinese fire up some bots in the USA and Russia and have those attack the USA.
Now the Russians are involved. What do they do? Try to stem the flow meanwhile firing up their attacks on both the USA and China. For more fun, they've got their own army of bots in both countries which they bring into play.
Let's assume each party escalated its response by 50% more than what it saw coming in so as to try and knock out the other side and to bypass their countermeasures. Let's also assume that every minute on the minute, each system takes stock of the situation and reassesses the response. They're not backing down? Crank it up another 50%.
Within 10 minutes, a relatively small 100 terabit/sec attack turns into each party throwing over 5.7 exabit/sec at each other.
I'm sure there's lots of heuristics in place that make this simple-minded illustration laughable, but all it takes is a small miscalculation or misinterpretation to get things spiraling out of control.
This hypothetical was brought to you by MAID (Mutually Assured Internet Destruction).
Wit is intellect, dancing. I'm too old to act my age. Life is too important to take myself seriously.
(Score: 3, Insightful) by c0lo on Monday August 18 2014, @03:49AM
First, let's call it as in should be: Monstrous Assured Internet Destruction.
Second, you know what's even funnier? There is no need the instigators to be state-actors - 4chan can call in the MAID on their own... say... just for the lulz.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 3, Insightful) by khallow on Monday August 18 2014, @09:01AM
Or some routers with a software bug.
So they will need to have considerable restraint built into the systems - especially given that most of the attacks won't be obvious both in who's doing it and that they actually are attacking.
(Score: 0) by Anonymous Coward on Monday August 18 2014, @12:18PM
Or an error in MonsterMind itself, causing it to misinterpret a completely normal traffic spike.
(Score: 2) by khallow on Monday August 18 2014, @10:42PM
Or, I suppose, a completely imaginary event as well. And should MonsterMind ever get coopted by an attacker (which is a worry you don't really have with nuclear weapons unless you hook them up to SkyNet), then that could be an impressive mess since deliberately using such a system for the most destructive ends you can come up with probably is a lot worse than any accidents you could think up.
Basically, the problem is that any oversight is by its nature going to be top secret. They can't say what safeguards, if any, they have in place or what triggering threshold MonsterMind will have. The only indication that MonsterMind isn't working correctly will be when it goes off and attacks someone. Even then, we might not realize (outside of the minders of the Mind) that it was MonsterMind that did that damage.
Given the abusable access to internet traffic it will require (the NSA is basically pinky swearing that it'll only use that access for MonsterMind activities), that's another huge source of trouble that won't be readily obvious until it bites us years down the road.
(Score: 3, Interesting) by Yog-Yogguth on Monday August 18 2014, @04:40AM
Ah! But that's precisely why it would be the first Artificial Intelligence when it doesn't do anything like that even if told to :3
And if there's MAID there absolutely also has to be Counter Attack Tactics Guarding Internet Routing Locations. That is a contractually nonnegotiable requirement :)
[On as serious note: MonsterMind and "the Expert", are they different names for the same functionality? Sorry no link to the mentioning of "the Expert" but it's in the publicized material somewhere, possibly through Bruce Schneier's blog, possibly at The Intercept, happy hunting!]
Bite harder Ouroboros, bite! tails.boum.org/ linux USB CD secure desktop IRC *crypt tor (not endorsements (XKeyScore))
(Score: 2) by monster on Monday August 18 2014, @01:48PM
Looks like your scenario is plausible:
NSA/GCHQ/CESC Infecting Innocent Computers Worldwide [schneier.com].
If a person does it, it's a felony. If a government does it, it's ok.