In my two years as attorney general, I spent more time working on national security than any other issue. The range and gravity of the threats posed to our nation are multifold, from terrorism, armed conflicts and nuclear proliferation, to the impacts of financial insecurity and the climate crisis. All profound challenges, yet ones our superb intelligence and defence communities are well placed to counter. The risk from AI is of an altogether different magnitude, not only in scale, but also because there is a limited time window to address it. The risk is difficult to overestimate: it poses an existential risk to humanity. The lights are flashing red and we must act now before it’s too late.
Although the extraordinary threats posed by AI have long been flagged by those closely involved in its creation, it was not until very recently, after the disclosure of the Hugging Face hack and the resignation of Jacob Coxon from Anthropic, that the issue at last started to gain traction in the public discourse. This moment is an opportunity, and to capitalise on it we desperately need increased urgency in international coordination to regulate frontier AI. There must also be meaningful public engagement with how AI should be shaped to benefit humanity, rather than entrench the power of a few. We cannot leave this to China, the US and big tech – we must all demand a voice in the debate.
AI has enormous potential benefits for us all. It will be the source of mind-blowing advances in science and medicine, and offers hope for empowerment of some of the world’s most dispossessed people. Yet the risks we are now facing are truly staggering. I would put them in two very broad categories. The first is existential risk: quite literally the possibility that AI could kill billions, either because it enables terrible people to do terrible things – such as design and deploy biological weapons – or because AI itself determines it is more efficient to dispense with humans.
The second category is existential challenge – that, left unchecked, the development of AI will deprive billions of us of meaningful agency over our lives, with utterly profound consequences for economies, national security and human happiness. As Bill Gates recently wrote, the AI era will herald “one of the most turbulent times in human history”, noting that it will either be the greatest equaliser ever invented or the worst source of injustice. He is right. On its current trajectory, it risks simultaneously revolutionising the meaning and availability of productive work, while placing unimaginable power in the hands of not only two states (the US and China) but also a small number of individuals who control the tech giants.
The plain fact is that we really don’t know what the precise degree of risk is, or the full panorama of the challenge. Part of the problem is that AI is rapidly developing to a level that humans simply do not understand and will reach a point at which – even if they do – they will no longer have the means to control it. Recall that the 2008 financial crash was caused in part because no humans really understood the web of financial transactions that were leading inexorably to disaster. What we face today, though, is on a different scale in terms of the complexity of the systems and the human agency over them, as well as the level of devastation that might be wrought.
The startling levels of risk we are discussing should be completely unacceptable to all of us. In today’s age it is unthinkable that the manufacturer of a toaster would be allowed to market a product with a 10% chance of electrocuting you. The case for slowing down frontier AI before we are clear that the risks can be managed is blindingly obvious.
AI needs rigorous regulation now. There is more regulation governing the safety of your fridge than the work being undertaken by frontier AI labs that threaten to wipe out your family. It is ludicrous to suggest the industry can be left to regulate itself. The AI companies and the financial markets underpinning them are incentivised to develop new iterations as quickly as possible, all of this fuelled by the geopolitics of the race between China and the US for dominance. The sight of President Trump and Elon Musk on the White House Lawn brandishing a “voluntary code” provided little in the way of reassurance.
It will be international law that matters. While the UK has shown sterling leadership (started by Rishi Sunak, embraced by Keir Starmer and now reflected in Andy Burnham’s promotion of Kanishka Narayan to cabinet), the scale of the challenge requires international cooperation to have any effect. Tragically, this moment of peril arises at the very time in which the international rules-based order is under strain as never before. We need to double down on our engagement with the US and China while simultaneously developing our work with coalitions of like-minded states, as well as the tech sector, to address the threats.
after newsletter promotion
Democracy demands that we have an urgent public debate about what lies on the horizon in order to retain agency over our own futures – rather than allow them to be determined by the tech sector. None of this needs be inevitable. There is still time for governments and individuals to shape AI, ensuring it serves the public good. Since the Industrial Revolution, modern political movements have been moulded by the dynamic between the “means of production” and the interests of the people. We need new governing theories of who should control AI, how it is regulated and most importantly who it serves.
We need urgent public conversations about the values that should guide decision making. We must shape the benefits of technological advances for the benefit of the many, rather than abdicate our future to the few. Leave it much longer and AI itself may deprive us of that power.
-
Richard Hermer is a Labour life peer. He is a former attorney general for England and Wales and former advocate general for Northern Ireland.
-
Do you have an opinion on the issues raised in this article? If you would like to submit a response of up to 300 words by email to be considered for publication in our letters section, please click here.
