Weaponized AI: Existential Threat Yesterday

This article was written for The European Sting by one of our passionate readers, Mr. Alexandros K. Liakopoulos. The opinions expressed within reflect only the writer’s views and not necessarily The European Sting’s position on the issue.

Current affairs bomb us all every second with hard doses of reality. Reality is not about conversations, bullet points and presentational remarks. It’s about facts. Fact of the days: AI is a reality, it happened and is now building up. Question of the day(s): is AI a future existential threat for humanity? No, it is an existential threat – period. To arrive to the Future(s), one needs to survive the Present and its sequence. The Present has become much more precarious due to the coming of AI. Nevertheless, this reality tends to escape us: a brief Bing/Google search under the terms “ai existential threat today” gives no direct link. With the help of Bing’s AI agent, the framework of past and current discussions emerges. Clearly, Bing knows what’s going on, can articulate it and can even joke about it: when I asked it AI is an existential threat today, it told me “sorry, I am not in the mood of discussing my mental state today ”. Seriously.

The coming of AI has multiple implications for the near and far future, some of which have been discussed and analyzed. The foreseen implications are a tiny fraction of the unforeseen ones, as we have seen in all previous technological revolutions. AI is a technological revolution so groundbreaking that will inevitably and quite quickly lead to a whole societal paradigm shift. The changes will be enormous, humans and societies will be challenged to their core and not everyone will be able to cope with the fast pace of change. Granted. All of the above are true. Except they have already happened and are happening as these lines are getting written. We live in a fundamentally different world and the fact we cannot recognize the fundamentality of the change is due to mental inertia: Homo Sapiens does not cope well with change, especially when it comes in very fast timeframes, or in big doses, if you prefer. Due to the coming of AI, Change comes in speeds we had never achieved in the past, in doses we have never digested, affecting more and more people, in wider and wider spaces, physical and digital, all around Time. What can go wrong with our perception, right? Well, everything.

AI is an existential threat today, each and every day its weaponization is not banned. AI’s weaponization is getting exploited already. This the existential danger we face and it is very real. Weaponized AI massively altered the strategic equilibrium of hard power at the planetary level and will do even more so quite fast. An arms race on weaponized AI is on, implicating the most ambitious and capable countries in the world, non-state actors and even individuals: all actors may become much more efficient and effective in delivering lethal blows to their enemies. Beyond legitimate and semi-legitimate actors, imagine a world were lone-wolves find their steroids: cheap, deadly and accurate weapons on their hands, enabling them to deliver repetitive strikes instead of just one, very efficiently, very effectively, very fast and precisely where they should be delivered to create the targeted results. This has happened and is happening..

Power as such has changed, due to the coming of AI. Weaponized AI creates havoc in Gaza, where Israel delivers the “smartest” war in the history of mankind, testing multiple technologies on the field to deliver lethal blows of precision. The technology promises to limit the number of civilian casualties. Does it do it? Well, who can tell?

Some indications exist that weaponized AI is also tested in Ukraine, or even in Moscow. This fundamentally changes the strategic environment the mankind so carefully and so painfully crafted in the 20th century, after two World Wars and a nuclear arms race that led to a stability of terror, aka Mutually Assured Destruction, which came to be known by its volumes-speaking acronym, MAD. MAD became possible and effective exactly because the “superpowers” had similar weapons, operated with similar processes and in similar speeds and were managed by humans, operating under their biological, psychological, physical, mental and ethical constraints. Said otherwise, all the factors of supremacy in the battlefield were common to the holders of the ultimate means of power of the past, making supremacy impossible for all and destruction assured to all, irrespectively to who had the first strike. MAD was valid because it led to No-Last-Man-Standing after the fight. 

All the above factors have been massively disturbed due to weaponized AI. MAD no longer applies, a point that was made by Dr. Kai-Fu Lee in his 2021 article in The Atlantic under the title The Third Revolution In Warfare: “The deterrence theory does not apply, because a surprise first attack may be untraceable… autonomous-weapon attacks can quickly trigger a response, and escalations can be very fast, potentially leading to nuclear war. The first attack may not even be triggered by a country but by terrorists or other non-state actors. This exacerbates the level of danger of autonomous weapons.” Dr. Kai-Fu Lee is not unfamiliar with the subject: currently he runs a multibillion fund in China, investing in AI.

Autonomous weapons based on open source AI can be cheap, built fast and achieve extraordinary results in precision, in speed and in stealth. They can make personalized killing a reality. Slaughterbots are no more science fiction; in 2018 they tried to kill the President of Venezuela Nicolas Maduro in a failed attack that injured seven people of his guard. Ukraine has built an army of drones, 100% decentralized and operated both individually and in concert. May 16th was the day Sanctuary AI presented its humanoid, Phoenix, and some hours later Tesla presented its own, Optimus. Technologically we are a decision away from deploying them in the battlefield, further disrupting the strategic equilibrium of the planet. Playing with the strategic parameters of the international system is as dangerous as it gets. 

New technologies transform the battlefield and the combatants: as some civilians may become combatants from distance, as a part time occupation, delivering “smart” asymmetrical blows to the enemy by their own means and under no control whatsoever, all civilians may be seen as legitimate targets. The new weapons can be much more precise, as they can use facial recognition, body language, voice and other clues to hunt, find, identify, target and execute any given person, group of persons, gender, or even ethnic groups. The targets will have already been previously located by other AI systems analyzing physical information, digital clues, radio frequencies and telephone conversations in order to place the target(s) on specific extended locations, no matter their size so long as they remain specific. This specificity barrier is still small, but it will inevitably cover the whole planet as our communication infrastructure progresses in parallel with AI development. 

Elon Musk’s X Starlink company, which builds a low-orbit internet satellite constellation while it also builds analogous infrastructure for the US Space Development Agency, resurrecting and boosting President Raegan’s Strategic Defense Initiative (SDI), aka Star Wars, has already weaponized, both willingly, by the US government collaboration with Musk’s enterprises, and unwillingly to them both, maybe, by Ukrainians of the Army of Drones. Retaliation has not yet come from Russia and may never come; yet, more than imaginable and conceivable, it gets highly probable that Russia will try to do so. After all, it has stated it considers all options and keeps its whole arsenal in elevated preparedness. China is on its way of building its own satellite constellation, as is also the EU. Star Wars on steroids, coming soon, but not in the cinemas: above our skies.

Further, it is less than two years since South Korea’s Agency for Defense Development (ADD) announced a breakthrough in stealth drone technology; and less than two weeks since Kremlin was hit by two drones. The last time Kremlin was hit was by Napoleon. Even to consider throwing a pebble at a window of the Kremlin and succeeding would be inconceivable until it happened, as it still is inconceivable for the White House. No no-stealth-drone technology could have managed to deliver the hit, no matter how imperceptible in terms of damage: the Kremlin was hit, this is the damage. The impossible, throwing even a stone at the Kremlin, became exponentially possible: not only a stone, but a spot-specific hit. Since the technology is already there, armies fighting are using it, no matter if armies are not so strictly defined anymore or if they cannot even be considered an army: amateur enthusiasts turn terrorists for a cause will proliferate, so long as the technology gets cheaper and easier to obtain.. 

Since the enabling infrastructure develops more and more and the systems become all the more compatible with each other, both their use and their scope will “explode”: intelligence explosion is not the problem, not for now at least; weaponized AI explosion is. More people using the infrastructure for malicious purposes either under management of a legitimate actor, like a state, or not, is the recipe for creating eternal havoc, an endless spiral of violence development, of all against all and any against any, with autonomous weapons and killer robots at the disposal of legitimate and non-legitimate actors. In fact, the very notion of legitimacy, legal responsibility, even legal personality of the actors has been shaken to the core with the coming of AI and are currently trying not to be scattered all around as the debris of yesterday. Who is a legitimate actor when weaponized AI is used by third parties to deliver blows to whoever they feel like? Who has the responsibility of the blow, what is the chain of command? All these questions, unanswered, pose the real dangers for the Now.

There is a widespread discussion on the ethics of AI, as sensitive and well-informed citizens are pushing their governments to establish solid barriers to the possibilities of AI, so it does not escape its black box and kill us all. Governments and international organizations have already published regulatory frameworks and continue to consult with the civil society to come up with the right solutions, in very brief time, while multiple other crises further deteriorate everyday stability. The ethical framework to build in AI and the legal framework to rule AI are undoubtedly cornerstone discussions. However, they have become slightly irrelevant to the Time in history we stand. We no longer are in 2015, when killer-robots were a future fear and intellectuals along with businessmen of all calibers, politicians of all beliefs, apostoles of all faiths and preachers of all dogmas came to sign petitions to stop them. We live in 2023. What was then an urge to avoid is today’s reality, worsened by orders of magnitude: killer-robots exist and fight, businessmen of great importance have become intrinsically connected to the warring machine of their nations, while they are on war, the political divides within countries of unequaled-in-the-past power no longer allow for greater consensuses, worldviews get massively replaced with echo-chambered realities and all these happen when there is the fastest and wildest transformation of the international system ever to happen in three generations. Can we really afford to discuss future threats?