Is the future of warfare already here, coded in lines of algorithms and humming in server farms? The recent revelation that the Pentagon branded Dario Amodei, CEO of Anthropic, a “liar with a god complex” isn’t about one company or one CEO. The real story here isn’t the clash of personalities – it’s the complete erosion of the line between civilian technology and military application. We’ve moved beyond “dual use” – the idea that a technology can be used for good or ill – to a point where the very design of modern AI inherently carries the potential for violence.
The panic surrounding Amodei and Anthropic stems from fears that their large language models could be weaponized, used for surveillance, or to manipulate populations. But this isn’t a novel concern. As Thomas Christian Bächle and Jascha Bareis argue, today’s AI isn’t simply capable of violence; it’s built on foundations that actively facilitate it. Consider the ubiquity of drones. They’re touted for delivering medical supplies to remote areas, for efficient logistics, even for offering a breathtaking new perspective through aerial photography. Yet, the same technology is readily adapted for reconnaissance, sabotage, and, tragically, lethal force – a reality starkly illustrated by their deployment in Ukraine, where they’ve disrupted civilian air travel and inflicted devastating damage. This isn’t a bug; it’s a feature of a system designed for adaptability and autonomy.
Source material: iai.tv.
This ambiguity isn’t unique to drones, of course. Every technology is shaped by, and in turn shapes, the social processes around it. But AI, with its universality and increasing autonomy, amplifies this dynamic. ChatGPT, for example, can be a helpful research assistant or a disturbingly convincing source of misinformation. That same open functionality is easily translated into war technologies, automating the identification, tracking, and even engagement of targets without meaningful human intervention. We’re not talking about science fiction anymore. We’re talking about a shift in the very nature of violence, one where our reliance on digital infrastructure creates vulnerabilities ranging from attacks on power grids to the manipulation of public opinion through “deep fakes.”
The “dual use” label, once a convenient shorthand for assessing risk, is now analytically useless. It implies a clear separation between civilian and military applications, a distinction that’s rapidly dissolving. The Haber-Bosch process, initially developed for fertilizer production, was quickly repurposed for creating explosives during World War I – a historical precedent for the blurring of lines. But the speed and scale of modern technological development, particularly in the realm of AI, are unprecedented. A 3D printer in your home can now produce functional firearms, demonstrating how easily intended uses can be subverted. The intended use and the actual use are diverging at an alarming rate.
The core of the problem lies in the fundamental nature of computation. It’s not developed for a specific purpose; it’s a universal problem-solving tool. Algorithms and data are universal representations of the world, and the rise of algorithmic processing has reinforced the idea that tasks can be automated without human agency. AI takes this a step further, with machine learning enabling systems to adapt and find their own solutions. This pursuit of autonomy, while promising in many fields, is deeply unsettling when applied to weaponry. Who is responsible when an autonomous weapon system makes a fatal error? How can we even trace the decision-making process? These questions remain largely unanswered.
The current fascination with autonomous weapon systems reflects a long-standing human desire for control, a quest to impose order on the chaos of war. AI is presented as a way to overcome human weaknesses, to achieve “clean” strikes with surgical precision. But this vision is deeply flawed. Targeting, the process of identifying and selecting targets, is inherently subjective and value-laden. Reducing individuals to data points and assigning them risk scores is a violation of fundamental human rights. And the liquidity of geopolitical violence, exemplified by drone warfare, means that anyone, anywhere, can become a target at any time.
This isn’t just a technological issue; it’s a psychological one. The swarming capabilities of drones, for example, evoke primal fears of being overwhelmed by an incomprehensible force – a modern echo of biblical plagues or Hitchcock’s The Birds. These displays of technological prowess are often used as a form of psychological warfare, signaling power and deterring potential adversaries. China’s recent drone swarm demonstration, recognized by Guinness World Records, is a prime example. It wasn’t just about the technology; it was about sending a message.
The pursuit of “ethical AI” – with concepts like transparency, fairness, and robustness – is a laudable goal, but it’s ultimately insufficient. Regulatory frameworks become meaningless when the political systems employing these technologies disregard human rights and ethical norms. There’s no effective global regulation of AI, either in the civilian or military domain. The realistic path forward isn’t a utopian vision of world peace or a comprehensive treaty banning autonomous weapons. It’s acknowledging the current geopolitical reality: a constant rivalry, a landscape of hybrid warfare, and the inevitability of technological escalation.
The new objective isn’t to eliminate conflict, but to maintain an equilibrium – a constant chase between attack and defense, malicious software and countermeasures. Just as cybersecurity isn’t about achieving absolute security, warfare isn’t about achieving absolute peace. It’s about managing risk and maintaining a balance of power. This may seem cynical, but it’s a pragmatic assessment of the current situation.
Watch for this: in the next 18 months, a major geopolitical power – likely China – will publicly demonstrate a fully autonomous drone swarm capable of operating without any human intervention in target selection. This won’t be a test; it will be a statement. And it will force the rest of the world to confront the uncomfortable truth that the rules of war have already changed, and there’s no going back.
