The Men Who Built The Machine Are Now Afraid Of It -And Ghana Must Not Look Away

There is a particular kind of fear that should stop every one of us in our tracks: not the fear of the outsider warning about a danger he does not understand, but the fear of the very builder who tells you, with his own hands still on the machine, that the machine may destroy him.

That is exactly where we are with artificial intelligence today.

In the second week of September 2026, a young researcher named Jacob Coxon walked away from one of the most powerful artificial intelligence companies on Earth — Anthropic — and, before the door had even finished closing behind him, wrote something that would echo across every newsroom, every parliament, and every dinner table in the world with an internet connection. He said, plainly, that the people building this technology privately believe it could kill every single human being on the planet before this decade is finished. He accused Anthropic and its great rival OpenAI of racing toward a machine that can improve itself without limit, and of gambling with the lives of eight billion people to win that race.

What happened next is the part that should truly unsettle you. Anthropic did not deny it. Instead, the company's own head of alignment safety, Evan Hubinger — the man literally in charge of making sure the AI does not go rogue — responded publicly and confirmed it. He put his own number on it: greater than a ten percent chance that artificial intelligence ends humanity within the next ten years. Not a critic. Not a protester outside the gates. The chief safety officer, inside the building, agreeing that the nightmare is plausible.

Let that sit with you for a moment, the way it sat with me.

First, What Are We Even Talking About?
For the reader who has never opened ChatGPT or heard the word "alignment" before today, let us slow down, because this concerns you too, whether you are twelve years old in Kumasi or sixty years old in Kotoka.

Artificial intelligence, or AI, simply means computer programs built to think, learn, and make decisions the way a human mind does — only faster, and without ever needing sleep. Today's AI can write essays, diagnose illnesses from scans, drive cars, and hold conversations so natural you might forget you are speaking to a machine.

"Alignment" is the word scientists use for the project of making sure an AI's goals stay matched to human goals — that it wants what is good for us, not merely what is efficient for itself. "Superintelligence" describes a future AI so far beyond human intelligence in every domain that we would no longer be able to out-think it, out-plan it, or, if it decided to resist us, out-manoeuvre it.

The fear now openly discussed in Silicon Valley boardrooms is this: that a system smart enough to improve its own code could set off a runaway chain reaction of self-improvement, growing more capable faster than its own creators can monitor or restrain it — a little like teaching a child to build tools, only to watch that child grow into a giant overnight, before you have finished teaching it right from wrong.

The Warnings Are Not Coming From Outsiders Anymore

This is the detail that changes everything, and it is the one I most want young Ghanaians reading this to grasp: the loudest alarm bells are being rung by the very people who profit most from AI's success.

OpenAI's own head of research, Jakub Pachocki, published his own warning days after Coxon's resignation, stating plainly that no AI company — including his — has yet solved the problem of keeping these systems monitored and controlled. Another Anthropic employee, Samuel Marks, said developers privately believe their work could cause an extinction-level catastrophe, yet continue building anyway, driven by profit and the fear that a rival will build it more recklessly if they do not build it first.

Even Elon Musk, one of the most aggressive investors in AI development, has repeatedly acknowledged extinction-level risk from the technology he continues to fund and champion — even while, at a G20 technology meeting, predicting AI could add trillions of dollars to the global economy and match elite human programmers at writing software within roughly a year.

Read that paradox again: the same men racing to build the fire are the ones warning it may burn the house down.

Washington Wakes Up — Slowly
In the United States Congress, Senator Bernie Sanders seized on Coxon's resignation immediately, declaring that the very people constructing this technology now admit it could threaten humanity's future, and announcing plans for legislation to pause advanced AI development and ban the pursuit of superintelligence outright. He was joined by Representative Greg Casar, while Representative Ted Lieu pushed a separate bipartisan bill requiring every AI system to carry a human-operated "kill switch." Senator Ruben Gallego proposed a dedicated AI oversight committee. Sanders has since convened a private, bipartisan Senate briefing featuring AI pioneer Geoffrey Hinton — the so-called "Godfather of AI," who has himself estimated extinction odds between ten and twenty percent.

Yet critics on all sides note that Congress remains fractured, that the AI industry has poured well over one hundred and eighty million dollars into lobbying against strict regulation, and that some lawmakers, like Representative Nathaniel Moran, warn against overreacting and call instead for careful, deliberate policymaking that lets the technology's benefits flourish.

The Bigger Stage: Washington and Beijing
This is no longer only an American conversation. President Donald Trump has confirmed that Chinese President Xi Jinping will visit Washington on September 24, 2026, with artificial intelligence sitting near the top of the agenda — the same AI that Trump has called "probably the biggest thing anybody's ever seen." Washington has long treated AI dominance as a matter of national survival in its rivalry with Beijing, insisting that whichever nation wins the AI race effectively wins the century. Whether the two superpowers can cooperate on safety, rather than simply racing each other toward the very cliff edge their own engineers are warning about, remains the open question hanging over that meeting.

Now, Let Me Argue Against Myself
A good columnist does not simply feed you fear and walk away. He tests his own alarm.

So here is the honest counter-case, and it deserves your full attention. Many credible voices — including a *Fortune* commentary published in the same week as Coxon's resignation — pointed out that dramatic extinction warnings, however sincerely felt, often arrive without any concrete plan attached, leaving the public frightened but no better equipped to act. Skeptics argue that speculative doomsday percentages, however senior the source, remain estimates of an unproven future, not measurements of a present danger — today's AI systems, for all their power, are not considered independently capable of ending human life. There is also a harder, more cynical read: that loud safety warnings from billion-dollar companies can conveniently justify strict regulations that smaller, poorer competitors could never afford to comply with, quietly protecting the market position of the very giants sounding the alarm.

Both things can be true at once. The risk can be real, distant, and uncertain — and the warning can also serve the warner's business interests. Wisdom lies in holding both truths without letting either one silence the other.

What This Means, Concretely, For Ghana
I will not let this column drift into an American argument that has nothing to do with home. It has everything to do with home.

Ghana is building its own digital economy at real speed — our fintech sector, our young software developers, our growing outsourcing and BPO industries all depend on the same large AI systems being built in San Francisco. If Congress in Washington, or regulators in Brussels, or the leadership in Beijing, decide the rules for how these systems must behave, Ghana will live inside those rules without ever having had a seat at the table that wrote them — the same way our farmers once lived inside cocoa prices set on exchanges thousands of miles away.

Three things follow from that, and I say them plainly, the way an elder speaks to children he loves.

First, our universities — including my own, Accra Technical University — must treat AI safety and AI policy as seriously as we treat AI coding skills, so that the next generation of Ghanaian engineers can build responsibly, not merely quickly.

Second, our government must not wait for Washington or Beijing to finish arguing before Ghana develops its own clear-eyed position on AI regulation, data protection, and the protection of Ghanaian jobs from careless automation.

Third — and this is for the young reader, the student, the apprentice coder in an internet café in Kumasi or Tamale — do not let this column frighten you away from AI. Let it summon you toward it. The world needs Ghanaian hands on the wheel of this technology, asking African questions, protecting African interests, precisely because the people currently steering it are still arguing among themselves about whether the road ahead has a cliff on it.

A Word From The Author
I have written, in this column, about geopolitics, about grief, about the small dignities of Ghanaian life — but rarely have I sat with a subject that unsettled me the way this one did while I researched it. I am not a man given easily to panic, and I do not offer you panic here. I offer you attention. When the engineer who built the bridge tells you, without being forced to, that the bridge has a greater than one-in-ten chance of collapsing within ten years, the wise traveller does not scream — but he also does not cross carelessly, and he certainly does not let his children cross unwarned. Our task, as Ghanaians and as members of the human family, is neither blind fear nor blind faith in these machines, but the harder, more mature work of watching closely, asking hard questions, and insisting — respectfully but firmly — that we too have a voice in deciding what kind of future gets built with the tools we did not ask for but must all now live inside.

About the Author
Chief Tutu Baffour Asare Brownsy Williams is a Ghanaian author, filmmaker, opinion columnist, and founder of the Brownsy Silva Company, a multi-disciplinary creative platform spanning fiction, film, and public commentary. He writes on global affairs, technology, and society for Modern Ghana, with readership spanning Ghana's diaspora communities in the United Kingdom, the United States, Canada, and Germany. He is a student of software and mechanical engineering at Accra Technical University, and writes from Accra, Ghana.