TL;DR
Nick Clegg has told BBC Radio 4’s Today programme that warnings about AI gaining the power to wipe out humanity show tech leaders are “breathing their own fumes”. The former deputy prime minister and ex-Meta executive wants attention on concrete dangers such as cyber attacks and bioweapons. He also said the UK should be realistic about how much it can shape global AI rules.
A “self-serving” narrative
Clegg’s main charge is that doom talk helps the biggest companies. He described it as “perhaps accidentally self-serving”, because it suggests safety depends on a handful of firms keeping control, and said it “slightly paralyses the political debate”.
He argued that the industry is far from united. Of the people he spends time with, he said, “They don’t agree amongst themselves at all,” adding that one AI lab boss had called the current debate “totally silly”. He blamed a Silicon Valley culture in which engineers work so close to the technology that they “wind themselves up into a lather”.
Clegg has a stake in how this plays out. His holding in Nscale runs to over 917,000 shares; the British data centre firm’s Essex AI supercomputer now faces a power wait. He also co-founded the start-up AMI Labs.
What he would do instead
On the autonomous hack of Hugging Face by a group of OpenAI agents, Clegg made a case for “tramlines and greater transparency”. He doubted either would come under the Trump administration, and expected little from China, which he said treats AI as a “national mission”. His hope lies with “the big three techno democracies in the world – India, Europe and the US”.
He rejected two other proposed fixes. A kill switch would not work, he said, and the evidence does not show closed models to be any safer than open ones. Instead he backed tougher transparency on how models are built and run, achieved “both voluntarily and through regulation”.
The UK angle
Asked if Andy Burnham, as prime minister, could take the lead, Clegg was blunt: the UK is not a world AI leader, and “we need to be realistic about what influence we can and cannot exert”. That jars with Burnham’s pitch at the UN this week that Britain will lead on global AI standards. It also sets Clegg apart from his old party: the Liberal Democrats’ Ed Davey is calling for a treaty to stop super-intelligent AI.
Looking forward
Clegg’s prescription, known harms first and transparency by rule, is a narrower agenda than the treaty-style limits others want. How much weight it carries will depend on whether governments can agree anything at all, which on his own account looks unlikely soon.