AI Arms Race: the Alarming State of Cyberwarfare c.2023

the AI Arms Race: Global Escalation

APOLOGIA: This post on the current state of the AI Arms Race is nowhere near complete. But it’s how I choose to publish. Honestly, these days, I have so many thoughts about AI in my head that I deem “ready to publish” that I could go 24/7 for 7 days straight and not be complete. So sometimes I write long screeds offline, editing them to perfection, and then… an amazing amount of time, they stay sitting there, in my local file system, never +++qqquuiiiiite+++ ready for publication. HOWEVER: If I just whip open the WP editor, and start typing, and hit “Publish” as soon as I get either tired or nervous that my browser is about to crash.. well.

Then, its published. And then, the pressure is on, because I can see, real time, that there are humans (and robots!) reading what I *just* prematurely published. And THAT… that energy, that concern, hypermotivates me to edit the hell out of the uncooked thought that I just published, and across the next 4 hours I pump out 4-8 revisions/edits of it that make it progressively better. And then I’m on to the next premature post, and it’s edit cycle. And then, 3-10 days later, after a bunch of people have already read it, I re-visit it, and re-read it, and realise that it still needs some tweaks and interconnects with other pages, and I tweak it, and massage it, and… across the months… it all comes to fruition.

So that’s a long way of describing my process, and why the post you’re about to read (and many others), while it has the *seed* of a thought, isn’t quite fully formed yet. But keep coming back. This (and many other posts) come slowly and surely to maturity and perfection. And at that point, I will remove this disclaimer. Boom.

The AI Arms Race

As with many of the posts here, the intent of this post is to shine some light upon critical happenings that you, the global Citizen, might be completely unaware of at the moment. Particularly, the state of what you may have only, up until now, thought of as the stuff of science fiction movies: this being, the present state of AI Cyberwarfare.

In this particular post, we’ll be talking about World War III… not the one fought with guns and missiles and blood. No, this one is far more sinister. This one is fought with ones and zeroes, bits and bytes, dollars and yen, information and… beliefs. And in the end, this present war will prove to be far more lethal than any that have preceded it.

Sign of the Times: US Cyber Command

from the official federal government website:

“United States Cyber Command (USCYBERCOM) is America’s unified combat command for the cyberspace domain, formed in 2010. It is headquartered alongside the vaunted and ultra-secretive NSA (National Security Agency) at Fort Meade, Maryland. USCYBERCOM is a top-level military unit that operates globally in real time against determined and capable adversaries, which take the forms of individual bad actors, international criminal and terrorist organizations, foreign and corrupt corporations operating against U.S. commercial & consumer interests, and antagonistic nation-states.

The Command comprises military, intelligence, and IT capabilities. Its mission is to direct, synchronize, and coordinate cyberspace planning and operations to defend and advance national interests in collaboration with domestic and international partners.

The Cyber Mission Force, first authorized in 2012, originally consisted of 133 teams, with a initial combined force of almost 6,200 military and civilian personnel.”

As of 2022, U.S. Cyber Command employs some 25,000 soldiers, who run 3 shifts 24/7/365 to monitor, warn, prevent and alert against enemy attacks coming in via domestic internet access points. These soldiers do not carry .45 sidearms to work. They carry laptops.

These soldiers do not wear helmets and bulletproof vests. They wear VR headsets and haptic hand controllers. These soldiers don’t call in for air support. They deploy AI counter-defenses to thwart an exponentially escalating, non-stop assault upon every available digitally-connected domestic target… which today would include your smartphone, your doorcam, your Alexa and “your” Tesla, in addition to your banking, credit card, and retirement accounts.

The initial US cyberwarfare annual budget (c.2012) was well north of a billion ($1B) dollars, and has increased by roughly $1 billion per year ever since, to the present-day budget proposal of $12+ billion for FY 2023.

A Few Quick Cyberwarfare Graphs

Some foundational metrics will help us frame this conversation:

  1. graph of number of cyber attacks being launched against US interests daily (intercepts at US-International NOC routers / points of entry into the domestic network):

The “IC3” mentioned above is the Internet Crime Complaint Center, a division of the U.S. Federal Government responsible for collecting consumer & corporate claims of cyberattacks and cybercrimes.


Additional Graphs: TBP

  1. budget expansion of US CyberCommand (2012-2023)
  2. headcount growth of US CyberCommand (2012-2023)
  3. growth in # of daily attacks per US cyber soldier.
    1. attacks grows exponentially
    2. soldier headcount grows linearly
    3. therefor, number of attacks managed per soldier increases constantly.
    4. conclusion: the only way that soldiers can drastically increase their ability to respond to a greater volume of attacks in less time is to for each soldier to employ a “force multiplier.”... some form of technology or training that enables them to handle 10x-100x the threats, in the same amount of time as via traditional means. The only quantitatively effective force amplifier, in this case, is AI-augmented defensive measures.
  4. Spam metrics:
    1. total daily spam messages, growth 2002-2022
    2. total amount of all email that is spam
    3. total amount of all internet traffic that is spam
  5. Internet of Things (“IoT”) in the context of cyberwarfare.
    1. how making all components of infrastructure, from light bulbs to thermostats to doorbells to traffic lights… all internet-connected and IP-addressable… how making this “smart”, also makes the same vulnerable.
  6. Primary Targets / Attack Surfaces:
    1. Consumer ID (identity theft)
    2. Consumer bank accounts / credit cards (financial fraud / theft)
    3. corporate databases (mass identity theft)
    4. commercial infrastructure (i.e. hospital databases / ransomware attacks)
    5. national infrastructure (power, telecom, water, etc – true warfare meant to cause social unrest, injury & death)
    6. governmental functions & databases (DoD, IRS, etc)

Conclusion: Moving Forward, its AI vs. AI

The only solution against the problem of ever-escalating and increasingly sophisticated cyber-threats…

…is to Deploy Defensive AIs
to counter
offensive AI-spawned
cyberwarfare swarm attacks.

And pretty soon, really, humans are simply:

out of the loop.

Its AI vs. AI.

First on the internets, and then, god save us all, on the earth, in the physical realm (tap here to see the current SoTA in humanoid soldier combat robotics)

…and so, the AI arms race begins.


4-star general Nakasone explains the AI Arms Race to Congress

4-star Army General Nakasone, head of U.S. CyberCommand, testifies before Congress regarding the rapidly escalating threat of “bad actor” nation-state A.I. cyber-attacks on vulnerable US assets & infrastructure. (April 5, 2022 : link)


The Problem with LAWS: Lethal Autonomous Weapons Systems

One of the very plausible and highly problematic endpoints of this line of development is when weaponized AI makes the leap from the purely digital world of the internets into our physical environment, workplaces, and homes. There are already massive fleets, soon to be swarms, of UAVs… “unmanned aerial vehicles.” Any of the more substantial of these (think Air Force Predators) have a human crew of approximately 6 controlling their movements and weapons launches. However, at this point, that human role is largely symbolic, and is completely unnecessary for both the operation of the aircraft and its various weapons systems. All it takes is the throwing of a switch, and the craft moves into full autonomous mode. The technical term for such a leap is LAWS, or “Lethal Autonomous Weapons Systems”

For a chillingly plausible visualisation about where these lines of thought eventually lead us, watch:

a short film developed by the Future of Life Institute / DUST Films


An inexpensive, mass-produced micro-drone contains fully autonomous-AI navigation, targeting, and triggering software, as well as a 3g explosive charge capable of ending a human’s life.


Swarms of thousands of the microdrones can be airdropped en masse from high above enemy territory to seek-and-destroy individual members of the target population.

“The Future is 100% Autonomous”

This “full-auto” function, it should be stated, works equally well on “piloted” aircraft (and cars, btw). None other than Secretary of the Navy Ray Mabus famously prophesied regarding the state-of-the-art F35 Joint Strike Fighter plane:

“the F-35 should be,
and almost certainly will be,
the last manned strike fighter aircraft
the U.S. Military
will ever buy or fly

USNI News — April 15, 2015