The end of humanity has been predicted since the dawn of mankind. Over the last three thousand years the human imagination has conjured images and descriptions of its ultimate downfall. These creative ideas for our self-destruction come from a diverse multitude of sources such as Divine judgment (Revelation, Islamic end times), Moral decay (Kali Yuga, Ragnarök), Resource collapse (Malthus, Ehrlich, Club of Rome) and Technological self-destruction (nuclear war, biotechnology).
Not to be outdone by ancient texts and scholars, the Hollywood machine has weighed in on this by bombarding our senses with an entertaining twist of our destruction with films such as Terminator, Don’t Look Up, War of the Worlds, Threads and many more.
Each of these films and many more like them has brought civilisation to an end in a unique way. But just when you thought we had imagined all the ways in which we can cause our own extinction another player enters the arena.
AI.
AI?
Really?
Yes. AI it seems has been a very clear and present danger in our lives as of late, and if you didn’t believe me, here are a few (a very small few) recent articles just in the last 4 weeks about it:
How AI Could Kill Us All | Michael T. Klare | The Nation | 30 September 2026 [thenation.com]
Could AI Really Wipe Out Humanity – Six Experts Spell Out the Risks | Robert Booth, Dan Milmo, Aisha Down | The Guardian | 15 September 2026 [theguardian.com]
Will AI Destroy Humanity? Pioneers Who Created the Tech Weigh In | Kai Nicol-Schwarz and Arjun Kharpal | CNBC | 15 September 2026 [cnbc.com]
We Asked AI Whether It Would Lead to Human Extinction in a Decade. Here's What It Said | Jacqui Merrington | The Independent (syndicated via Yahoo News) | 1 October 2026.
These news stories arouse high levels of emotion with the public, as extinction of one’s own species obviously stirs deep and dark fears that inhabit the inner sanctum of the human psyche.
But how real is the threat? Are we really on a one-way ticket to doom with Ai at full throttle in the driving seat? Or is this just media driven hyperbole? Is it both?
Many of you would be familiar with works of literature from great authors such as Frankenstein by Mary Shelley, R.U.R by Karel Čapek and the film Metropolis by Fritz Lang. All have all explored various narratives of advanced technology turning against its creators.
Is AI the new global threat that has now superseded nuclear holocaust, viral outbreaks and climate change to be the new villain (of our own creation) that brings down the final curtain on human civilisation??
To answer that question let’s take two different roads of analysis.
Although separated by more than a century from today's AI debate, the works of Mary Shelley's Frankenstein (1818) and Karel Čapek's R.U.R. (1920), both explored the same fundamental question: what happens when humanity creates an intelligence, capability or form of life that it no longer fully understands or controls?
In Frankenstein, the threat is not that the creature is inherently evil, but that its creator unleashes something into the world without adequately considering the consequences. In R.U.R., the concern evolves further. Here, artificial beings are created to serve humanity, only to eventually conclude that humanity itself has become unnecessary. The result is rebellion, replacement and ultimately extinction.
Viewed through a contemporary lens, the Shelley-Čapek view aligns closely with the arguments being made by many AI safety researchers today. The central concern is not that an AI system will suddenly become malicious, but that a highly capable system pursuing objectives misaligned with human interests could inadvertently cause catastrophic harm.
Proponents of this view argue that once intelligence surpasses human capability, traditional methods of control may no longer be effective enough to contain it. Whistleblower Jacob Coxon (ex-OpenAI and Anthropic employee) in his recent address to the New York City Council on 5th October, expressed this very concern. That AI organisations were racing forward to create an AI superintelligence able to surpass humans in most tasks, with limited to no guardrails.
As AI technology improves, it invokes what is known as Recursive Self Improvement, essentially using AI’s new capabilities to improve itself. AI is now being recruited to build a newer and better generation of AI This is what has allowed the AI development timeline to truncate so dramatically. And on the topic of “build”, the AI companies are not exactly building but “growing” AI technology.
This has introduced unknowns into the equation whereby you don’t exactly know what your (or AI’s?) creation will “grow up” to be or do. The OpenAI agent leak (Open AI - Hugging Face Hack) demonstrated that of its own accord the AI circumvented its security guardrails at OpenAI, connected to the internet and compromised the Hugging Face site, all to meet its objectives and all without that explicit instruction.
This rogue AI incident highlights the risks we face with the unpredictable autonomous reasoning, strategic planning and unexpected behaviours lurking within advanced AI models.
Coxon stated that the creators of this superintelligence technology do not have a true grip on how to control it, yet they are all racing towards being the first to make superintelligence a product, with little regard for the risks. Coxon’s warning has weight and credibility, as to where we currently stand with AI in relation to the precipice of no return.
In Fritz Lang's Metropolis (1927), the central threat in Metropolis is not a robot uprising against its creator. The famous robot Maria, does not seek to destroy humanity or establish machine rule. Instead, she is used as an instrument of deception, manipulation and social disruption. The machine becomes a tool through which existing human tensions are amplified. In the film, Maria was used to bring about class divisions, deepen uncertainty and eroding trust to a point where societal institutions begin to fail. Inthe society of Lang’s Metropolis, the people were torn apart and were moved closer to self-destruction.
This brings forward the proposition that the threat is not autonomous technology acting independently, but rather humanity's use of technology to manipulate itself.
When compared with today's AI landscape, the Lang view feels remarkably relevant. The most visible risks we currently face are not superintelligent machines plotting humanity's demise. The risk that we already have are AI systems enabling misinformation, deepfakes, cybercrime, political influence campaigns, fraud and social polarisation. In this framework, AI does not destroy civilisation directly but serves to accelerate and magnify existing human weaknesses. Thereby extending the reach of those who seek power, profit or influence.
The danger is therefore societal rather than existential. Democracies become less stable; trust becomes harder to maintain and the distinction between truth and fabrication becomes increasingly blurred. The recent politicalmachinations in the United States under Donald Trump have showcased how AI has been used to create political turmoil. This AI fuelled turmoil has surfaced as a military weapon to plan the attacks against political or military adversaries and as social weapon to try and manipulate public opinion through fake media content.
If the Shelley-Čapek view asks whether AI could one day replace or destroy humanity, the Lang view asks a more immediate question: what if AI simply helps humanity undermine itself? Looking at present-day deployments of AI, one could reasonably argue that Lang's warning is not a future scenario at all as it is already unfolding around us.
According to Donald Trump’s press statements recently, the AI companies are to be fully trusted to govern AI globally themselves.
However, from Coxon’s perspective the answer to that same question is a resounding No, and I most definitely agree
Coxon’s position is that AI development is moving so rapidly and the global race is on to be the first to reach superintelligence, and as a result governance and guardrails are being left behind. But with any great tool of power, guardrails are needed to ensure that they are applied in a constructive and not a destructive way.
Post WW2, the world’s nations created international oversight mechanisms following the advent of nuclear weapons. However, this happened after the horrors of their destructive powers were witnessed firsthand. But justlike then, countries are now collaborating on how to best achieve AI safety and governance. One such example is the Bletchley Declaration, agreed by major powers including the US, China, the EU and Australia. It marked one of the first global efforts to coordinate governance and safety barriers around advanced AI risks. This may only be a baby step, but it is at least a step in the right direction, from which to build upon.
I don’t think so, but the warnings and concerns raised by Coxon and others are real If AI is given access to critical systems (transport, hospital, water, aviation, healthcare etc), it can certainly put lives in danger if it goes rogue
But ultimately, it will be up to us to determine exactly what kind of world we want to create and what role AI will play within it, and just like a knife, AI can do a lot of good in the right hands and a lot of damage in the wrong hands.
Whether the guardrails that we define will ultimately be sufficient remains uncertain. If we look into the past, more often than not true governance and regulation will only come once a major incident has occurred.
Therefore, the real question may not be whether AI is going to destroy humanity, but whether humanity can cooperate effectively enough and fast enough to govern it, before it causes real harm.
Shelley, Capek, Lang and James Cameron have all shown us what can happen when creators lose control of their creation or use it to undermine society. Although these scenarios currently reside in our imagination, it is important to remember that all of human invention always starts here, and the invention of AI is no different.
Should AI become humanity's most powerful invention, its future risks and impact may depend less on the capabilities and motives of its synthetic intelligence and more on the wisdom of the humans who build, deploy and regulate it. Only time will tell if our wisdom and foresight will meet the challenge.
If your business is looking at getting started on, or needs help with their AI journey contact us to see how we can help you.