A MARTÍNEZ, HOST:
President Trump says tech companies can self-regulate the risks of artificial intelligence.
MICHEL MARTIN, HOST:
And he signed an executive order officially dubbing the technology super intelligence for government agencies. This came after the president met with dozens of top leaders in the AI field on Tuesday, the same day he also rolled out an AI-driven website, America.gov, to help people navigate government services.
MARTÍNEZ: NPR's White House correspondent Emily Feng is with us to explain this. So, Emily, what happened at this meeting?
EMILY FENG, BYLINE: So there was no press in the meeting, but Trump came out afterwards, and he told the press how people had agreed companies could self-regulate on AI. He argued if companies break the law, we have the Department of Justice or the FBI that could rein them in. Though, to put a finer point on that, if that were to happen, the U.S. would need to put AI laws on the books, and that did not happen at this lunch. Instead, Trump posted a one-page document that CEOs from six of the big AI companies that were there signed. And this agreement says companies would broadly monitor their models and make sure they do not hack into other systems, but it offered no details on how this would work.
(SOUNDBITE OF ARCHIVED RECORDING)
PRESIDENT DONALD TRUMP: They don't want anything to go wrong. I mean, their companies are at stake. If something goes wrong, their companies are at stake. They're not going to let that happen. This is going to be all for the good, and we're going to have it be nice and safe. And they're really going to be policing each other.
FENG: Asked if this was enforceable, Trump said it was, quote, "morally binding." And so the question still remains whether powerful private companies can be trusted to follow through in perpetuity.
MARTÍNEZ: Well, it's morally binding. That might be enough. Now, which companies were participating in this meeting?
FENG: There were about 30 tech executives present. And I want to note the people who were seated right next to Trump during this meeting. They were Nvidia's Jensen Huang, Elon Musk and also CEOs from Meta, chip designer AMD - their CEO, Lisa Su, is actually the only female executive invited - as well as people from Amazon and Google. And notably, there were two representatives from Anthropic at that meeting, and they included co-founder Dario Amodei, who Trump does not have the best relationship with. If you remember, the Pentagon has effectively banned Anthropic. They've called it a supply chain risk, which came after the company objected to the potential use of its products for autonomous weapons.
And Amodei has been one of the most outspoken entrepreneurs for more AI regulation in the last couple of weeks, and he was notably excluded from a state dinner with Chinese leader Xi Jinping last week, though his competitors from OpenAI were there. And I'll note that Anthropic is a financial supporter of NPR.
Now, while Trump and Speaker Mike Johnson, who was also at this meeting, spoke glowingly about this agreement afterwards - they're calling it the AI Accords now - Amodei struck a more cautious note, and he suggested the mechanisms agreed upon in this document are still very much a work in progress. Listen here.
(SOUNDBITE OF ARCHIVED RECORDING)
DARIO AMODEI: The mechanism, how we address those risks, is still under discussion, but we all need to work we all need to work together to make sure that we can win and we can win safely.
MARTÍNEZ: All right. So we mentioned this AI-powered tool that Trump launched. What exactly is it?
FENG: It is powered by artificial intelligence companies Grok and Google's Gemini. And the Trump administration is really pushing it as an example of how AI can streamline and improve government. It's kind of like a concierge tool. You know, I tried it out. It's a chat function. You pull it up, you input questions on how to, say, renew your passport or enroll in Medicaid. You know, I was curious about whether it stores your data and whether it uses it to train the chatbot behind the tool. The model says it does not, and a senior administration official who is not authorized to speak on the record confirmed that your data is not being used to train Grok or Gemini.
MARTÍNEZ: All right. That's NPR's Emily Feng. Emily, thanks.
FENG: Thank you. 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.
