King Charles warns of AI existential risk at Dumfries House summit

The summit brought together figures whose companies are otherwise competitors: Nvidia, Google DeepMind, OpenAI and Anthropic all sent representatives to Dumfries House, and the royal household's announcement named only the companies, not the individuals. Bloomberg reported that Anthropic's global affairs head, Tino Cuéllar, attended and met the King privately, among several one-on-one meetings.

The messages from the attendees varied in emphasis. Jensen Huang focused on safety as a precondition for releasing products and on the value of open models that let researchers, universities, startups and countries build applications no single actor could imagine alone. Demis Hassabis put a timeframe on the debate, saying AGI will probably arrive in a few years with ten times the impact of the Industrial Revolution, and that the chance of something going wrong is definitely not zero. OpenAI's CFO, Sarah Fryer, said no single company or government can handle these questions alone.

What the gathering ultimately changes is likely to hinge on whether such statements turn into commitments the attendees are held to, and on whether the open-model protections Huang called for are formalized. For now, the summit's main output was the alignment of language among the executives present, not a specific measure.