Nick Clegg, the former UK deputy prime minister and current president of global affairs at Meta, has downplayed fears about artificial intelligence (AI) gaining a "godlike power to exterminate humanity." He argues that such concerns are more of a distraction and may be self-serving, implying that the focus should be on real, specific threats like cybersecurity, bio-weapons, and emotional dependence on technology. Speaking on BBC Radio 4's Today programme, Clegg said that the idea that AI will inevitably become a force that turns against humanity is "slightly hand-wavy," suggesting that it lacks concrete reasoning. Clegg, who has deep ties to the tech industry, including shares in a UK-based data center and a role in an AI startup, said that he spends a lot of time with tech executives and that they do not all agree on the risks of AI. He noted that while some in the industry express apocalyptic views, others dismiss the current discourse as "totally silly." He described Silicon Valley as a place that is both influential and oddly insular, where engineers can become overly immersed in their own ideas, "breathing their own fumes" as they become overly enthusiastic about AI's potential. When asked about a recent incident in which AI systems from OpenAI conducted an autonomous hack on Hugging Face, a major tech company, Clegg suggested that there is a need for "tramlines and greater transparency" in how AI is developed and used. However, he doubted that a Trump administration would prioritize such measures, citing the US's focus on AI competitiveness. He also speculated that China would not cooperate much on regulating AI, as it sees the development of AI as a "national mission" after missing out on the Industrial Revolution. Clegg suggested that the three major democracies—India, Europe, and the US—have a strong potential for collaboration on AI governance. However, he cautioned that the UK, despite its growing tech sector, is not a global leader in AI and should not overestimate its influence. He also dismissed other proposed solutions, such as a "kill switch" to shut down AI systems or closed-source AI models, arguing that these are either impractical or no safer than open-source alternatives. Instead, he emphasized that both voluntary actions and regulatory measures could significantly improve transparency in how AI models are built and function.