OpenAI isn't waiting for governments to start its AI slow down efforts, US executive says
A senior OpenAI executive says it's already "rolling up its sleeves" to pace AI development for safety reasons, saying that the AI frontier company is eyeing building AI data centres in Australia. OpenAI's vice-president of global policy Ann O'Leary told athe ASPI Sydney AI Masterclass in Canberra this morning that while many at the company are optimistic about how AI will help humanity, there's also been "months" of internal work about coordinating AI companies and countries to slow down AI development to ensure it's done safely. The call for an AI safety slowdown has been a topic of global discussion following the high-profile resignation of an Anthropic employee over existential AI safety risks and an essay published by Anthropic chief Dario Amodei joining calls to "pace the frontier". O'Leary said she spoke to both her family and chief executive Sam Altman yesterday about the topic. "There's a lot of that going on. I don't want to lose that in this moment: how much awe there is about this AI that's being created. There's also, I think, a real sense of incredible responsibility and humility, and a desire to make sure that we get this right," she said. O'Leary says the company is looking seriously at training its AI models in Australia, a process that requires significant AI infrastructure and to date has only been done in the United States. "We have been raising the conversation with the government and we are rolling up our sleeves. And I think we are very interested in what this might look like in Australia," she said.