Malwarfare: Triton, the first Malware designed to kill people

[Amaroq_Starwind]

https://www.technologyreview.com/s/613054/cybersecurity-critical-infrastructure-triton-malware/

This is the first known Malware specifically designed to kill people, as far as I can tell. I feel like there couldn't be a better time for somebody to take a more offensive approach to cybersecurity...

[Hyperwolf122]

I'm just now hearing of this, thats insane, I didn't ever think such of a thing would exist?

[Amaroq_Starwind]

It was inevitable. Some groups are just filled with horrible, horrible people.

[Hyperwolf122]

Yeah I agree, that's awful 

[exile360]

This is only the beginning.  Once true AI comes online we'll see automated attacks the likes of which the world has never known which discover and exploit 0-hour vulnerabilities in real-time and build threats and attacks on the fly to counter any security measures and crash through any barricades in their path with startling speed and efficiency with both brute force and machine learning heuristics as well as (potentially) evolutionary cognitive adaptation to develop and push their malicious code and have their way with devices and systems, short-circuiting anything not equally equipped with 'good' AI to counter the threats (and even then, a race condition or 'tie' between such systems could easily result in a drastic drain on resources and infrastructure resulting in DoS of varying scope anywhere from the endpoint under attack to parts of the internet's infrastructure as the machines use distributed networks to pool resources and bring 'troops' into the battle to attack one another).

It sounds like sci-fi, but much of what we've seen over the past decade was once thought impossible in the field of computing and technology, and I assure you, this will happen.  It's only a matter of time.

[Amaroq_Starwind]

There's a system called MAYHEM, designed to hunt down vulnerabilities zero-day vulnerabilities and correct them on the fly. Combine that with a hybrid Digital/Analog/Quantum supercomputing platform, and you might have an effective defense. I know for a fact that it can be done.

[exile360]

That's exactly the kind of tech I'm talking about (though much earlier in development/dumber compared to the kind of true AI I was referring to).  The issue is, that once it comes down to AI vs AI everyone loses because then it's just like a conflict between two AVs on the same system; resource usage escalates and eventually the system crashes or freezes due to a race condition.  Imagine that on a scale that impacts entire sectors of the internet's backbone/infrastructure.  We'd live in a world at constant risk of being attacked and/or used by AIs for their botnets and roaming blackouts of major portions of the web on a regular basis potentially becoming an everyday occurrence.  Not a pretty sight.

[Amaroq_Starwind]

1. https://www.wired.com/2016/06/demonically-clever-backdoor-hides-inside-computer-chip/ 
	2. https://en.wikipedia.org/wiki/Hybrid_computer 
	3. https://en.wikipedia.org/wiki/Analog_computer#Resurgence 
	4. https://www.dwavesys.com/take-leap 
	5. https://spectrum.ieee.org/computing/software/mayhem-the-machine-that-finds-software-vulnerabilities-then-patches-them  

Here we go... Hopefully something good will actually come out of me sharing all this information.

[exile360]

Yep, I was already familiar with most of that, but again, I'm referring to real, true, actual "thinking" AI, not just complex math and branching algorithms (which is literally all that exists right now).  Think of it like a super genius controlling one or many machines trying to hack into a network while another genius of equal measure attempts to thwart their efforts.  Eventually a stalemate will occur as the system(s) and/or infrastructure being used/attacked run out of hardware resources to feed their efforts so ultimately in the end you get a successful Denial Of Service (DoS) attack whether that was the intent or not.

As for current AI techniques (which aren't true Artificial Intelligence at all, just complex algorithms/math, not real 'thought' in any sense of the word):

https://www.techrepublic.com/article/ai-vs-ai-new-algorithm-automatically-bypasses-your-best-cybersecurity-defenses/
https://securityaffairs.co/wordpress/75206/malware/deeplocker-ai-powered-malware.html
https://techbeacon.com/security/how-ai-will-help-fight-against-malware
https://blog.trendmicro.com/artificial-intelligence-could-remake-cyber-security-and-malware/
https://gizmodo.com/the-malware-of-the-future-will-have-ai-superpowers-1830678865
https://securityintelligence.com/anti-i-the-future-of-ai-generated-malware/
https://www.techrepublic.com/article/using-ai-enhanced-malware-researchers-disrupt-algorithms-used-in-antimalware/
https://www.dailydot.com/debug/ai-malware/

Edited March 9, 2019 by exile360

[Amaroq_Starwind]

This is why the security providers need to get their hands on the quantum computers first, and why we need to start using mesh nets.

[Amaroq_Starwind]

Just out of curiosity: would analog computers be more resistant to Malware?

[exile360]

15 minutes ago, Amaroq_Starwind said:

This is why the security providers need to get their hands on the quantum computers first, and why we need to start using mesh nets.

It doesn't matter who gets what 'first' because in the end the other side will always catch up and likely eventually leap ahead at some point resulting in a stalemate or loss (especially since the bad guys are ALWAYS better financed than the good guys, even the largest of good guys like Microsoft and Symantec, as the bad guys are financed by rogue governments, wealthy criminals and organized crime).

13 minutes ago, Amaroq_Starwind said:

Just out of curiosity: would analog computers be more resistant to Malware?

No idea, not my area of expertise.  But I'd imagine any device can be hacked/made to run 'bad code' (the original hackers used analog phones to hack systems/networks through audio tones over phone lines).