King speaks to AI leaders about the importance of safety
'We need sufficient means of control before it is all too late'
King Charles has told leading AI figures, including Nvidia's Jensen Huang and Google DeepMind's Demis Hassabis, that the technology needs safeguards before it is too late.
Speaking during a summit at Dumfries House in Cumnock, Scotland, the King acknowledged that AI has the potential to improve and save lives, but also noted recent warnings about the threat it could pose.
"There seems urgency in adequately considering the existential dangers of such technologies falling into the wrong hands, and being used in potentially catastrophic ways," the King said, according to Reuters.
"Surely we need sufficient means of control before it is all too late?"
Nvidia CEO Jensen Huang, whose company has profited hugely from the AI boom, is among those who this week rejected calls for a slowdown in the pace of AI development.
He instead favours self-regulation, arguing that companies shouldn't (but stopped short of saying "wouldn't") release models they believe to be unsafe.
"You should go as fast as you can, but no faster than that," Huang told reporters at the Scottish summit.
He added that the King had been "one of the first [figures]" to speak to him about AI safety.
Other guests included OpenAI CFO Sarah Friar, Anthropic's Chief Global Affairs Officer Tino Cuéllar, papal adviser Paolo Benanti, and Minister for Artificial Intelligence Kanishka Narayan.
Narayan said the event had focused on the responsibilities companies had to ensure they were building AI for people's interests, not solely their own.
He also said the meeting would not be a one-off, but part of a "series of conversations" about the UK's role in responsible AI development.
Computing says:
Rivals OpenAI and Anthropic both coming out at the same time and saying they need to slow down has the same tenor as their responses to the agentic AI attacks this summer: more like a marketing exercise than true concern for safety – although if it results in the right outcomes for the wrong reasons, we won't argue.
I had an interesting conversation with delegates at our IT Leaders Summit yesterday, who pointed out that open-source systems are generally about 70% of the way towards matching frontier models' capabilities within a few months of release. If that's the case, and companies can get most of these capabilities for free, how tempting could an OpenAI or Anthropic IPO really be for big retail investors?
The delegates felt that announcements like this, and the warnings over AI dangers earlier this year, are part of a long game to bump up share prices more than any real concern for safety.
Good to see that it isn't only journalists keeping cynicism alive and well.