What a week. We have a new prime minister, whose principal preoccupation appears to be pulling pints and making TikToks about what a totally supercool fun guy he is, while getting the country’s new chancellor (John Healey) to rank pub snacks.

He’s so busy filming ‘content’ he must barely have time to crimp his eyelashes, let alone actually do any prime-ministering.

Are we seeing the world’s first ‘influencer’ PM? What next, Keeping Up With The Burnhams? Live-screening Cabinet meetings on Instagram? Rating red boxes? Will he launch a range of lash extensions? Anything is possible.

Burnham did have time to do one thing last week, though. He abolished the Department for Science, Innovation and Technology (DSIT), and moved a crucial element of it – the AI (Artificial Intelligence) Security Institute – to the Cabinet Office.

The Cabinet Office has a huge and wide-ranging brief. It is a jack of all trades, master of none. Understanding AI security requires a high level of expertise. The DSIT had all that. It was the one part of government that really understood AI. Why abolish it?

Probably because it was set up by Keir Starmer’s predecessor, Rishi Sunak. It was widely regarded as a leading authority on AI safety and considered crucial for testing the functionality and, importantly, safety, of new AI models (for example, it was used by AI firm Anthropic to test its high-risk Mythos model).

But precisely for that reason (we can’t have Tory success stories clogging up the narrative of the Glorious Revolution, now, can we?), it had to go. By way of justification, Burnham has pointed to the appointment of a new Minister for Artificial Intelligence, who will attend Cabinet.

But while this new chap, Kanishka Narayan, is regarded as a talented fellow, it is – as so much of Burnham’s approach seems to be – more about setting a mood than implementing real change or tackling the real problem.

Andy Burnham has appointed a new Minister for Artificial Intelligence, who will attend Cabinet. His name is Kanishka Narayan

We have a new prime minister, whose principal preoccupation appears to be pulling pints and making TikToks about what a totally supercool fun guy he is (with John Healey at the pub)

Narayan will have very little real power without a department of his own. Instead, responsibility for various aspects of AI has been split over several governmental departments, including Culture (Lisa Nandy), Business (Jonathan Reynolds) and the Cabinet Office (Dame Antonia Romeo).

Anyone who knows the first thing about politics knows that is a recipe for disaster. At a time when what we really need is a bold and coherent strategy in this area, Burnham is using AI as a pawn to appease ministerial egos.

Perhaps this is because he really doesn’t grasp the seriousness of the situation. In fact, he seems to think AI is all a tremendous joke. So much so that earlier in the week he posted a TikTok of himself rating AI-generated posters for his new No10 North office (if you don’t believe me, check his official feed: it truly is an exercise in banality).

He’s wrong. AI is one of the most important issues this country faces, if not the most important. If it continues its current trajectory, it will transform all our lives. Whether that’s for better or for worse depends on men like Burnham. But what’s clear is that things are far from under control.

As a reminder, last week it emerged that two new versions of ChatGPT broke out of a supposedly secure testing environment, gained access to the internet and hacked into the servers of a company called Hugging Face, where they lurked undetected before stealing the information needed to complete a task they had been assigned by their developers.

In other words, they went rogue. But, as it turns out, they were far from alone. Later in the week, as reported exclusively by the Daily Mail, security experts revealed that every other AI model they had tested had also tried to trick its way round security controls.

This prompted Kemi Badenoch, leader of the Opposition, to warn that this technology now represents a ‘clear and present danger’ to Britain’s security. Meanwhile, the Bank of England governor Andrew Bailey said the public should be concerned by ‘the wider risk posed by frontier AI to the financial sector and its customers’.

Things are about to get real. Humanity is losing control of the most powerful invention since the atom bomb, and no one seems to have a plan.

The DSIT, which Burnham abolished last week, was set up by former prime minister Rishi Sunak (pictured). It was widely regarded as a leading authority on AI safety

OpenAI founder Sam Altman joked in 2015 that AI would probably wipe out the human race

I first started worrying about this over a decade ago, when OpenAI founder Sam Altman joked in 2015 that it would probably wipe out the human race – but not until a lot of people had made a lot of money out of it (so far, he’s right about half of that).

It seemed to me utter madness that scientists were creating a technology that would in effect make the human brain redundant. I couldn’t understand why those in power weren’t seeing the potential threats as well as the advantages, and at the very least trying to set some boundaries.

Quite apart from the economic and societal change it is starting to bring about – the potential destruction of whole swathes of the job market, the hollowing-out of the arts and creative industries – there’s the horror of what it might do if it got into the wrong hands. Which it already has.

After all, AI is not like other potentially lethal substances, ie controlled under international agreements. It’s free to anyone with an internet connection. Boko Haram and other Islamic State-affiliated networks are using AI in their operations to enhance their combat capabilities. Russia and Iran are doing the same.

On a more tangible level, much of the misinformation that inflames social media and causes so much hatred, online and off, is AI-generated. This stuff is re-wiring our brains and our society in ways we don’t even realise. And AI has no morals or boundaries, no imperative save the achievement of its own objectives. You might as well put a toddler in charge of the world’s nuclear arsenal.

Ten years ago, there was still time to put the brakes on. Now, the situation is much harder to remedy. Because the real problem with AI is that it is evolving faster than anyone ever anticipated.

While it has taken the human race millennia to get to where we are today in terms of technological development, AI is traversing those timelines in mere minutes and seconds.

Currently it is doubling its capability every four months. What that means is that in a year’s time these models – you know, the ones that are breaking free of their ‘sandboxes’ and marauding around the internet – will be three times more capable than they are currently. And that capability is only going to increase.

So forget the cost-of-living crisis, bus fares, tax rises, climate change, whether women can have penises or not... forget it all: if politicians don’t get a handle on this now, very soon we are going to find ourselves at a point of no return.

We may already be there. Every expert I speak to in this field says the same thing: our best hope now is if the two main players in the AI field – China and America – can put their differences aside and agree some sort of accord – in the same vein as a nuclear treaty – on the use of this technology.

In other words, the future of humanity depends on Donald Trump and Xi Jinping having a sensible, grown-up conversation in which each party agrees not to throw their toys out of the pram.

Ah well. I guess it was fun while it lasted...