Humanity Has Seen This Pattern Before. It Ends in Blood

From Innovation to Weaponization: A Recurring Pattern

In the early days of commercial drone technology, people used them to film weddings.

Photographers discovered they could capture aerial footage of remarkable beauty. Engineers used them to inspect bridges and power lines without sending workers into danger. Humanitarian organisations delivered medicines to remote communities. The technology was celebrated. It was clean, creative, and full of possibility. Articles were written about the golden future of drone innovation.

Today, drones kill people. Not occasionally. Systematically. Drone warfare has become one of the defining features of every major armed conflict on earth. Autonomous and semi-autonomous systems carry lethal payloads across international borders. They select targets based on algorithmic processing. They execute strikes without the human deliberation that every existing framework of military ethics was designed to require. The technology that filmed weddings is now the technology that ends them.

To be clear, today's drones are not fully autonomous in most cases; the lesson is not about autonomy alone, but about how quickly a technology celebrated for civilian benefit can become central to the machinery of bloodbath.

The same trajectory played out with social media. With the internet. With nuclear technology. With every sufficiently powerful tool humanity has developed: the creative applications come first, the destructive ones follow, and the governance frameworks arrive — when they arrive at all — decades after the damage has been done and the deaths have been counted.

Artificial Intelligence and the Governance Window

Artificial intelligence may become the most consequential technology humanity has ever created. And we are, again, in the early phase; celebrating the creative applications, marveling at the productivity gains, writing articles about the golden future; while the window for shaping its trajectory toward human benefit rather than human catastrophe closes quietly around us.

The governance window is not infinite. It is not even comfortable. The technologies that define the next era of warfare, surveillance, and social control are being built right now; without frameworks, without treaties, and without accountability.

Why This Time Is Different
I want to be precise about what makes this different from every previous technology governance failure. Every prior dangerous technology had a limiting factor. Nuclear weapons required fissile material that was extraordinarily difficult to produce and has been tightly, if imperfectly, controlled. Biological and chemical weapons required specialized laboratory infrastructure and expertise. The barriers to entry were high enough that, in most cases, the capability remained concentrated in the hands of state actors operating within at least some framework of accountability.

Advanced artificial intelligence has no equivalent barrier. The marginal cost of intelligence; of the capacity to plan, to optimize, to deceive, to coordinate action toward objectives — is approaching zero. The same architectural principles that produce a medical diagnostic system can, with modest adaptation, produce a system designed to identify and neutralize human targets. The same optimization capabilities that power recommendation algorithms can be directed at goals that are not aligned with human welfare. And the same language models that explain biology and write poetry can generate disinformation at a scale and quality that no detection system currently in existence can reliably counter.

We are building minds we do not yet know how to control, and deploying systems whose objectives we cannot reliably specify in ways that remain safe as those systems grow more capable. Many leading researchers believe we are moving toward artificial general intelligence; systems that can outperform humans across all cognitive domains, without having solved the alignment problem. The race between capability and governance is not close. Capability is winning, and widening.

Private investment in artificial intelligence reached well over $100 billion globally in recent years, while international governance frameworks remain fragmented, voluntary, and largely unenforceable. Every year that passes without serious multilateral governance progress is a year in which capabilities advance, deployment widens, and the eventual reckoning grows harder to survive with the institutional order intact.

Three Measures That Can Still Alter the Trajectory

Three things are required with genuine urgency, and I will state them plainly because the diplomatic language of international policy forums has so far produced approximately nothing.

First, an international treaty framework for AI development analogous to the nuclear non-proliferation treaty. Not an aspirational declaration. A binding, verifiable commitment; with real consequences for violation on the development of the most dangerous AI capabilities, their deployment in weapons systems, and their use against civilian populations. This does not require stifling commercial AI development. It requires the application of the same institutional seriousness humanity managed to achieve for nuclear technology to a technology that is, by every measure, more broadly accessible and more rapidly advancing.

Second, a global moratorium on autonomous lethal systems; AI-guided weapons that can make the decision to kill without meaningful human oversight. The drone has shown us the arc. The AI-enabled autonomous weapons system will take that arc to its logical conclusion. Every month that passes without an international prohibition on autonomous lethal systems is a month in which the technical capability advances and the diplomatic window narrows.

Third, the genuine inclusion of every world region in every AI governance forum. Not as observers. Not as courtesy invitees. As co-architects of the frameworks under which the technology will be deployed globally. AI governance designed in Silicon Valley and Brussels, for Silicon Valley and Brussels, will serve Silicon Valley and Brussels. The Global South's governments, researchers, institutions, and communities must be at the table where the rules are written; not presented with the rules once they are finalised.

The Politics of Delay
I want to address, directly, the objection I always hear at this point: that governance will slow innovation, that competitive dynamics make unilateral restraint impossible, that the genie is already out of the bottle. These objections share a common structure: they treat the governance failure as inevitable rather than chosen.

Every year without serious multilateral AI governance is a choice. It is a choice made by governments with the institutional capacity to lead and the political will to avoid the short-term costs of constraint. It is a choice that future generations will inherit, in consequences that no diplomatic communiqué will be able to soften.

The Closing Window
The window is closing. This is not a metaphor. The technologies that define the next era of warfare, surveillance, social control, and economic disruption are being built right now. The institutions capable of governing them are not. Every year that gap persists is a year in which the eventual price of governance failure; paid in displacement, in conflict, in the erosion of the democratic systems that protect human freedom, grows larger.

We have been here before. With every powerful technology that humanity has failed to govern before deploying at scale, the pattern is the same: the creative phase, the celebration, the governance gap, and eventually the reckoning.

The question is not whether the reckoning is coming. It is whether we will choose to govern now, while the window is still open, or whether we will wait — again — to pay later. In blood!

Emmanuel Ezeoka is the Founder of the World Organisation for Human Advancement. He writes from Accra, Ghana.