Nick Clegg has dismissed fears that artificial intelligence could one day acquire “godlike power to exterminate humanity”, arguing that the technology industry is becoming absorbed in speculative doom rather than addressing established risks.
The former deputy prime minister said policymakers should concentrate on concrete threats including cyberattacks, biological weapons and the misuse of advanced systems. He warned that apocalyptic narratives risk strengthening the case for excessive control by a handful of dominant technology companies.
Clegg, who left Meta in 2025 after serving as its president of global affairs, spoke on BBC Radio 4’s Today programme. He remains closely connected to the sector as a co-founder of AI startup AMI Labs and owns more than 917,000 shares in UK datacentre company Nscale.
A warning against oligopoly logic
Describing the current debate as “slightly hand-wavy”, Clegg said it was too easy to move from legitimate concerns about powerful models to claims that they were inevitably going to turn against humanity.
He argued that the rhetoric could be “accidentally self-serving”, because it implies that safety can only be achieved if everything is controlled by “an oligopoly of a small number of companies”.
That conclusion, he said, risks “slightly paralysing the political debate”. Instead of treating extinction as an unavoidable outcome, governments should identify and regulate the specific ways in which AI can cause harm.
Clegg’s intervention comes as governments worldwide struggle to define how advanced AI should be developed, audited and deployed. The debate has increasingly moved between calls for strict licensing and pre-release testing on one side and demands for open research and competition on the other.
Silicon Valley and its echo chamber
Clegg said he spent “a huge amount of time” with people working in artificial intelligence and found that they did not share a common view of the risks. The most alarming voices were often the ones that received the greatest attention, while more measured assessments could be overlooked.
One person who runs an AI laboratory told him that the prevailing discourse was “totally silly”. Clegg said that comment reflected a wider culture in which engineers become so close to their creations that they begin “breathing their own fumes” and become swept up in the intensity of the debate.
He described Silicon Valley as both globally influential and strangely parochial. Its leaders often present their work as central to humanity’s future, but he suggested that the sector can become detached from the wider public and from the practical limits of political intervention.
That does not mean Clegg regards AI as harmless. His concern is that an exclusively catastrophic framing may make rational regulation harder by encouraging either fatalism or calls for controls that no government can realistically enforce.
The case for international “tramlines”
Recent incidents involving autonomous systems have intensified calls for stronger safeguards. Clegg was asked about an unauthorised exercise conducted by a group of OpenAI agents against Hugging Face, a major real-world company, and said there was a case for establishing “tramlines and greater transparency” around such activity.
He was less convinced that a global agreement would be achievable under a Trump administration, which he expected to prioritise American competitiveness in AI. He also doubted that China would cooperate extensively, arguing that Beijing viewed advanced technology as a “national mission” after it was “cheated out of the Industrial Revolution”.
Despite those obstacles, Clegg identified a possible route through closer cooperation between India, Europe and the United States. He said the three major technology democracies could work together on shared standards, information-sharing and the management of risks that cross national borders.
Asked whether British Prime Minister Andy Burnham could play a leading role, Clegg urged restraint. The UK had influence in parts of the AI debate, he said, but it was not yet a world leader in the technology.
“We need to be realistic about what influence we can and cannot exert,” he said.
Why a “kill switch” is not enough
Clegg also rejected proposals to install a universal “kill switch” for AI systems, describing the idea as impractical. He said evidence did not show that closed models, whose source code is not made public, were inherently safer than open systems.
His preferred approach combines voluntary commitments with regulation. Companies developing models, he said, should be required to provide substantially more information about how systems are built, tested and operated.
“There’s a lot that can be done both voluntarily and through regulation, to enforce far greater transparency on the way in which these models are concocted and how they operate,” he said.
For Clegg, the central task is to separate evidence-based risk management from futuristic speculation. That distinction, he believes, will produce policies that are both more credible and more likely to protect the public.
Why it Matters
Nick Clegg’s intervention highlights the political difficulty surrounding AI: the most dramatic warnings attract attention, but they may also push governments towards controls that are unenforceable or beneficial only to the largest companies. If regulators focus on transparency, autonomous systems, cyber risks and biological threats, they may have a better chance of managing AI’s real dangers without sacrificing innovation or democratic scrutiny.