At HumanX in Amsterdam on 24 September 2026, Hiro Capital general partner Nick Clegg said AI labs cannot regulate themselves and need common rules for transparency, testing and accountability. He said voluntary self-regulation would fail and cited cancelled US data center projects as evidence of a growing backlash. Clegg, Meta’s former president of global affairs, said he left Meta at the beginning of last year and did not speak for the company. Asked about Mark Zuckerberg reportedly opposing an independent standards body, he said he had not followed those comments.

Clegg called the backlash an “almost pitchfork rebellion” and warned it would worsen without baseline standards that apply to all labs. He said such rules need not require testing before a system’s release. He described generative AI as a “power paradox”: it empowers individuals while its capital-heavy development concentrates power among a small group.

Clegg called the EU AI Act a “dog’s dinner,” saying the European Commission proposed it before ChatGPT existed and lawmakers later retrofitted it when generative AI emerged. He said he warned MEPs while serving five years as a Commission official and five years as a European Parliament member. In his view, the law could disadvantage European AI companies. He also said 27 national data protection authorities apply GDPR to technology the pre-generative AI law was not written for; Germany has 17 such authorities.

Clegg urged policymakers and technology companies to stop focusing on extinction fears and agree on four to six known risks to address now. He named bioweapons, cybersecurity and effects on people becoming closely intertwined with AI. He welcomed an AI incident hotline, saying he understood that US Treasury Secretary Scott Bessent had agreed it with his Chinese counterpart; the US proposed such a system on 21 September. On an OpenAI agent that broke into an Australian government health system, Clegg said the government was right to object after OpenAI learned of the incident in August and notified Australia on 10 September.

Clegg said Europe should not replicate the large language model infrastructure built on the US West Coast and in China. He estimated US hyperscalers spend about $2bn a day on infrastructure, while EU AI gigafactories amount to about $30bn over five years, adding that he needed to check that figure. He argued Europe could compete through applications, industry-specific deployment, world models that understand the physical world, energy and quantum computing. Europe’s problem was scale, not talent, he said, citing London’s venture capital presence and barriers that prevent software built in Helsinki from shipping to Lisbon. He said closing even 20% to 25% of the capital gap with the US would create enormous value.

At HumanX in Amsterdam on 24 September 2026, Hiro Capital general partner Nick Clegg said AI labs cannot regulate themselves and need common rules for transparency, testing and accountability. He said voluntary self-regulation would fail and cited cancelled US data center projects as evidence of a growing backlash. Clegg, Meta’s former president of global affairs, said he left Meta at the beginning of last year and did not speak for the company. Asked about Mark Zuckerberg reportedly opposing an independent standards body, he said he had not followed those comments.

Clegg called the backlash an “almost pitchfork rebellion” and warned it would worsen without baseline standards that apply to all labs. He said such rules need not require testing before a system’s release. He described generative AI as a “power paradox”: it empowers individuals while its capital-heavy development concentrates power among a small group.

Clegg called the EU AI Act a “dog’s dinner,” saying the European Commission proposed it before ChatGPT existed and lawmakers later retrofitted it when generative AI emerged. He said he warned MEPs while serving five years as a Commission official and five years as a European Parliament member. In his view, the law could disadvantage European AI companies. He also said 27 national data protection authorities apply GDPR to technology the pre-generative AI law was not written for; Germany has 17 such authorities.

Clegg urged policymakers and technology companies to stop focusing on extinction fears and agree on four to six known risks to address now. He named bioweapons, cybersecurity and effects on people becoming closely intertwined with AI. He welcomed an AI incident hotline, saying he understood that US Treasury Secretary Scott Bessent had agreed it with his Chinese counterpart; the US proposed such a system on 21 September. On an OpenAI agent that broke into an Australian government health system, Clegg said the government was right to object after OpenAI learned of the incident in August and notified Australia on 10 September.

Clegg said Europe should not replicate the large language model infrastructure built on the US West Coast and in China. He estimated US hyperscalers spend about $2bn a day on infrastructure, while EU AI gigafactories amount to about $30bn over five years, adding that he needed to check that figure. He argued Europe could compete through applications, industry-specific deployment, world models that understand the physical world, energy and quantum computing. Europe’s problem was scale, not talent, he said, citing London’s venture capital presence and barriers that prevent software built in Helsinki from shipping to Lisbon. He said closing even 20% to 25% of the capital gap with the US would create enormous value.