AI debate reaches fever pitch, but the world remains unprepared

The global debate over artificial intelligence has hit a fever pitch. Apocalyptic warnings fill front pages one day, dismissals of the whole thing as hype fill them the next. Amid the noise, separating real danger from manufactured drama is becoming nearly impossible.

From Nobel laureate and AI pioneer Geoffrey Hinton to Silicon Valley poster boys such as Sam Altman and Elon Musk, some of the most prominent voices in AI have warned that the technology is becoming too powerful and needs stronger regulation before development goes much further.

But sceptics point to the uncomfortable economics behind the AI boom. Front-runners such as OpenAI and Anthropic are running up huge bills developing and operating increasingly powerful models, while promised revenues remain uncertain. With both companies eyeing stock-market listings, critics suspect that increasingly dramatic claims about AI’s power are simply aimed at keeping the venture capital flowing.

A Rattling Resignation

The latest cycle of the debate did not start in a boardroom or with the release of a new model: it started with a resignation post.

On September 8, AI researcher Jacob Coxon announced he was leaving Anthropic, warning that the industry was “racing straight to self-improving superintelligence and gambling with our lives”. The post reportedly drew more than 150 million views within 36 hours – a resignation letter moving faster than most governments can hold a hearing.

Four days later, Anthropic CEO Dario Amodei published “We Must Pace the Frontier,” calling for the industry to slow down and build safety checks between model releases. He stopped short of asking anyone to halt AI development, but the message was clear: even the companies at the centre of the race think it is moving too fast.

Then came the real anomaly: Sam Altman and Elon Musk, usually too busy sniping at each other to agree on anything, both said Amodei was right. When Silicon Valley rivals begin agreeing to hit the brakes, it is worth paying attention.

Strange Bedfellows

The unexpected alliances did not stop in Silicon Valley. In Washington, Democratic Senator Bernie Sanders and right-wing ideologue Steve Bannon shared a stage at a rally demanding that powerful AI systems be slowed down and regulated before new models are released.

That alliance should make anyone sit up. It suggests the AI debate is no longer confined to a familiar argument between tech enthusiasts and sceptics. People from opposite ends of the political spectrum agree that something must be done about the speed and power of AI.

Yet just as a domestic consensus forms, a stronger force pulls in the opposite direction: the geopolitical race between the US and China.

US President Donald Trump rejected any American slowdown, framing hesitation as a gift to Beijing and reducing the safety debate to a single social media post: the only guardrail AI needs is a “STRONG AND SMART (High IQ!) PRESIDENT”. Not sure if even the allies of the US would be comfortable with that assurance.

China Pushes Back

Beijing’s approach to the American safety push is more calculated. Chinese commentary likened Amodei’s proposal to a car in front slamming its brakes – not because the road is dangerous, but to stop the car behind from overtaking.

Combined with US chip export restrictions and rules against copying American models, the invitation to slow down looks like a demand for surrender dressed up as caution.

Meanwhile, China is hardly dismissing the risks. State Security Minister Chen Yixin warned that AI’s growing ability to find system flaws and generate malware threatens political and infrastructure security.

The central contradiction is inescapable. Most sides do not dispute that advanced AI can cause serious damage; what nobody agrees on is who has to slow down first, or who gets to write the rules.

The Global South Left Out

The result is a race where everyone says AI is becoming too powerful while believing that slowing down first is an even bigger mistake. But what happens to the countries neither setting the pace nor writing the rules?

While Washington and Beijing argue over who holds the wheel, the nations that will absorb much of the impact have barely been offered a seat at the table. India is one of them, as are the nations that broadly make up the Global South.

The consequences of AI will not stop at the borders of tech superpowers. They will reshape jobs, education, economies, cybersecurity, and public systems worldwide.

Yet at the BRICS summit in New Delhi, when Narendra Modi and Xi Jinping both raised the need for AI governance not dictated by a single bloc, domestic media discussions seemed more keen on the summit's vegetarian menu and officials snacking on nuts.

If that is the level of attention given when non-Western powers discuss AI's future, claims that the world is preparing for what comes next cannot be taken seriously.

An Echo of February 2020

Scepticism toward AI predictions is understandable; past hype cycles like blockchain and the dot-com boom failed to live up to their initial promises. But the fog around AI is not simply typical hype-versus-truth static.

Cybersecurity researchers are already warning about systems generating malware, economists are wrestling with a predicted job disruption, warnings of a stock market bubble are getting louder and financial markets are trying to price a technology whose ultimate impact remains deeply uncertain.

And beyond all that are the more extreme claims – that increasingly capable AI systems could eventually become able to improve themselves, take control of critical systems or act in ways that humans cannot reliably constrain. Some of those claims may prove exaggerated. Some may turn out to be wrong altogether.

Time magazine recently flagged an alarming comparison: what if we are slipping into the same complacency that preceded COVID-19? It reports that a growing number of AI researchers now admit that this moment feels uncomfortably like February 2020 – the last calm weeks before the world changed all at once.

Comparing a pandemic to an emerging technology is, of course, apples and oranges. But there is an unsettling similarity in our lack of preparedness.

COVID-19 proved how quickly a crisis can outrun global institutions, but its early days were defined by quiet complacency. Many had assumed it would be contained like SARS, bird flu, or Ebola while social media and even some public commentary offered everything from spicy food to alcohol as supposed protection.

It also showed that the countries making the biggest decisions were not necessarily the ones that could control the consequences.

AI is not COVID-19. But if the warnings about its risks prove even partly right, the world’s current state of unreadiness will look eerily familiar.