Threat of AI controlling nuclear weapons to be debated by UN Security Council during UK presidency
The potential threat of nations linking artificial intelligence to nuclear weapons, leaving computers to decide when to launch an attack, will be discussed by the United Nations Security Council later this month.
With concerns growing about AI one day posing existential risks to humanity, the UK is using its presidency of the Security Council for the duration of July to hold an international debate about how development of the technology could be regulated to ensure it is used safely.
Dame Barbara Woodward, the UK’s ambassador to the UN, said on Monday that “humanity stands on the precipice of this gigantic technological leap forward,” but added that serious “implications for international peace and security” must also be considered.
“There have to be concerns about how AI might be used, for example in nuclear technology – whether any country with nuclear weapons would consider handing over the management of those weapons to AI and how that could go wrong – and the same with automated weapons as well,” Dame Barbara told a press conference at the UN headquarters.
She underlined that the technology could have huge upsides in global affairs – supporting international development and humanitarian aid, as well as peacekeeping and conflict prevention.
“It could potentially help us close the gap between developing countries and developed countries… There is enormous potential,” she said.
But she added: “It doesn’t take too much to imagine what would happen if AI was unregulated in autonomous weapons, or what would happen if AI was put in control of nuclear weapons.”
Speaking later to i in an exclusive interview, Dame Barbara said the historic opportunities and challenges of AI are the reason why the UK is “very excited to be bringing AI to the Security Council for the first time” while chairing the body.
The UK Foreign Secretary James Cleverly will lead a meeting about the technology on 18 July with briefings from international computing experts and the UN Secretary-General, Antonio Guterres. This precedes a global summit on AI due to be held in London later this year.
The debate is “very timely,” said Dame Barbara. “It is a really important discussion and exactly what the UN Security Council needs to be doing as guardians and guarantors of international peace and security.”
She added: “I don’t think we need to be alarmist. I think we need to have a sensible measured conversation about it… I don’t think we can be too certain about doomsday scenarios.”
But there is a sense of urgency because “AI is developing so fast”, she said – acknowledging the way that ChatGPT, the AI-powered chatbot that was publicly launched last year, had “really hit us all”. She admitted: “We’ve played around with it here in our own mission at the UN.”
Mr Guterres has backed the idea of a new organisation like the International Atomic Energy Agency – which inspects nuclear facilities around the world – to ensure that AI is used safely. Dame Barbara said the UK is not convinced that the same model would work in this case, however, and that other methods of regulation should be analysed.
She said there were some fair comparisons to be made between AI and nuclear energy, which is beneficial when used to produce electricity but can also be used to create weapons. “We need to look at how we can de-risk as much as we can, and then use the benefits.”
But she recognises that AI could be a much more difficult genie to control once it is out of the bottle, because artificial intelligence could be secretly developed and used anywhere in the world, whereas nuclear weapons require mammoth physical engineering efforts that are hard to hide and can only realistically be run by governments.
