(SOUNDBITE OF ARCHIVED RECORDING)
PRESIDENT DONALD TRUMP: We can put guardrails. We can do this and that, but I think you have a lot of negative forces that are bringing it up that shouldn't be bringing it up. And they're bringing up things that won't happen.
ELISSA NADWORNY, HOST:
The remark seemed focused on keeping America competitive with China, rather than on regulation from Washington, dismissing rising fears after Anthropic CEO Dario Amodei published an essay on Saturday calling for international cooperation to slow down the development of AI. Sam Altman, the CEO of rival OpenAI, joined in on social media saying he agreed with Amodei and that OpenAI is delaying their plans for an expected blockbuster public listing. Jacob Coxon, a researcher who resigned last week from Anthropic, came on this program recently with a caution about the dangers of artificial intelligence.
(SOUNDBITE OF ARCHIVED NPR CONTENT)
JACOB COXON: There's the possibility we create something that, if it wanted to, could hack into any device on the planet, could use novel biological research to go far beyond what current scientists are capable of, could control, like, every robot in the world simultaneously.
NADWORNY: Coxon whirled (ph) about - worried about sleepwalking into a science fiction scenario. But will the growing backlash lead to a slowdown of AI development or any new government regulation? NPR's Shannon Bond joins us to discuss this weekend's developments. Shannon, Amodei and Altman, they run two of the companies behind the most capable AI systems. Why are they so worried about what they're building?
SHANNON BOND, BYLINE: Yeah, so Dario Amodei says he believes in the benefits of AI still, but he feels the technology is advancing too quickly now for safety to keep up. And so he's urging companies and governments to do what he calls pace the frontier, basically to collaboratively slow down development so they can address these risks. Here's what he told CNN.
(SOUNDBITE OF ARCHIVED RECORDING)
DARIO AMODEI: At the end of the day, the industry has to work together, and it's harder, but to some extent, the world has to work together.
BOND: And as you said, Altman quickly hopped on board, saying that he agreed. And in an interview with Fortune magazine, Altman also said that OpenAI would delay its highly anticipated initial public offering, which the company had been targeting to happen this year.
NADWORNY: So we've heard warnings about powerful AI for years. Why are these CEOs saying now is the time to take action?
BOND: That's true, especially from Dario Amodei. He left OpenAI, in fact, to co-found Anthropic specifically to take a more safety-focused approach to building AI. But listen, lately these fears have become inflamed about how quickly AI is advancing towards the point where it might no longer be governable by humans. You know, when Coxon, that Anthropic researcher, quit, he said both Anthropic and OpenAI are, quote, "gambling with our lives." They're not acting responsibly, and many in this industry agree with him. And it really seems, Elissa, we hit this tipping point recently with this incident this summer, where OpenAI systems went rogue and hacked a company called Hugging Face and hacked OpenAI itself. And Amodei pointed out to this hack and to other security breaches, including at Anthropic, in his essay, calling for these measures.
NADWORNY: So what are these AI leaders proposing to do?
BOND: Well, Amodei says that labs need to bring in independent evaluators who can look at safety. He says Anthropic is going ahead, taking that step now. Altman says OpenAI will do the same. And other AI leaders, including Elon Musk, Microsoft's Satya Nadella, Google's Demis Hassabis, they also chimed in this weekend, largely agreeing, though they didn't make specific commitments. OpenAI already said last month that it was temporarily slowing down development of its leaning (ph) edge AI models to make sure something like this Hugging Face hack does not happen again. But Dario Amodei says addressing safety risks is not something that these companies or the industry can do alone, and it's going to require global cooperation.
NADWORNY: What would global cooperation or global governance look like?
BOND: That is a good question. So what Amodei is calling for is for governments and industry to work together to create rules of the road, such as establishing common safety standards for AI and verifying compliance. But so far, we have not seen progress on that front. Officials from the U.S. and China are expected to meet this month to talk about AI safety, and AI is on the agenda at the U.N. General Assembly. So even if President Trump is brushing these worries aside, it seems like other world leaders are paying attention.
NADWORNY: Anthropic is a financial supporter of NPR, and we cover it like any other company. We've been talking with NPR's Shannon Bond, who covers how technology, politics, power and influence all intersect. Thank you, Shannon.
BOND: Thanks, Elissa. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
