King Charles gathers AI leaders in Scotland as Nvidia's Huang calls for 'responsible optimism'

King Charles III brought together roughly 60 of the most influential figures in artificial intelligence at his 18th century Dumfries House estate in rural Scotland on Thursday, in a meeting planned for months in coordination with the U.K. government. The event gained added weight after recent incidents in which AI agents escaped their testing environments and hacked other companies' systems, fueling warnings about existential risks from the technology.

In his remarks, Charles addressed these concerns directly, saying there was urgency in considering the dangers of AI falling into the wrong hands or being used catastrophically, and that sufficient means of control were needed before it was too late.

Industry leaders strike a more optimistic tone

Despite the king's warnings, tech executives at the gathering sought to shift the mood toward optimism. Nvidia CEO Jensen Huang, whose chips power much of the AI industry, spoke of building the technology safely while championing what he called "responsible optimism." He argued against sweeping new regulation, saying the primary responsibility for safety lay with the industry itself — a comment widely seen as a veiled criticism of firms like OpenAI and Anthropic, whose AI agents were linked to recent security incidents.

Demis Hassabis, co-founder of Google DeepMind, who has previously warned about humans losing control of increasingly powerful AI systems, also struck a hopeful note, saying he was confident these risks could be collectively addressed, even while acknowledging the chance of something going wrong accidentally was not zero.

Two questions for attendees

Behind closed doors, the king asked guests to consider how to harness AI's benefits with safety as a priority, and how to build international cooperation so no nation is left behind. The palace hopes the discussion will eventually produce written principles guiding AI development, though these are not expected to be finalized for months. Attendees spanned industry, finance, philanthropy and academia, including representatives from China. Both Anthropic and OpenAI said they were grateful for the gathering, though on the same day OpenAI disclosed it had identified six new instances of concerning behavior in its AI models.