Anthropic Calls for Global Pause in AI Development

SPEAKER_1 (0:00)

It's time to get Brex AF, a gen tech finance that eliminates manual work and puts you in control. Learn more at brex.com/af.

Luke Vargas (0:09)

Anthropic urges a global pause in AI development. Plus, the US sanctions Cuba's president as Washington further tightens the screws on Havana, and Taiwan's opposition leader tours America, promoting peace with Beijing.

Yoko Kubota (0:25)

Her visit will have some impact, potentially, on U.S.-China relations. Basically, how her close ties with Xi is going to play out in the US. And how Xi performs during this visit could have implications for that relations.

Luke Vargas (0:41)

It's Friday, June 5th. I'm Luke Vargas for The Wall Street Journal, and here is the AM Edition of What's News, the top headlines and business stories moving your world today.

We begin with breaking news in Washington. After a 19-hour session, the Senate has passed a $70 billion immigration enforcement bill to fund border patrol and ICE through the end of President Trump's second term without reining in a controversial anti-weaponization fund.

SPEAKER_2 (1:08)

On this vote, the ayes are 52, the nays are 47 This bill, as amended, is passed.

Luke Vargas (1:15)

Lawmakers from both parties had initially objected to the fund over concerns that taxpayer money could be used to compensate Trump's allies and reward those who assaulted police officers on January 6 But Republican senators eventually stopped short of using their political leverage to kill the fund, with only Alaska's Lisa Murkowski joining Democrats to vote against the bill. The House is expected to take up the immigration enforcement measure next week.

Anthropic is calling on top AI labs to consider slowing down their development, saying that AI systems are advancing so rapidly they may soon be able to improve themselves without humans. Our tech reporter Sam Schechner has more.

Sam Schechner (1:56)

What they're concerned about is that we may not be far from recursive self-improvement. That's a step where AI is able to effectively improve itself without human intervention. And that's a waypoint on the way to what some in Silicon Valley describe as the singularity, the moment when machines become more capable than humans.

Luke Vargas (2:16)

Sam said that warning, which was written by the head of Anthropic's Internal Research Institute, and was accompanied by a proposal for a global agreement on how to slow AI development, comes as the company pursues an IPO and recently concluded a fundraising round that valued it at nearly a trillion dollars.

Sam Schechner (2:31)

Anthropic has long been a company that's emphasized safety. I mean, that's sort of why it was founded, that they're concerned about the risks that AI could pose. And so that duality of we're pushing forward and we're concerned is kind of baked into what the company does in its messaging. There's some people who oppose that. They call it an agenda of regulatory capture, to basically say, let's bring in the regulators and protect our technological lead.

Others say that warnings like Anthropics are effectively a marketing ploy. Look at the godlike power of our tools, fear them, and buy them. Anthropic would say that they take safety seriously. And what they're doing is they're trying to start discussion now when there's a little bit of time left. But they're worried that it's getting close to too late.

Luke Vargas (3:20)

Meanwhile, we are reporting that senior US officials have discussed having the federal government take stakes in major AI companies after OpenAI CEO Sam Altman pitched the idea last year. Such a move could allow the government to share in the potential economic upside of the fast-evolving technology and give AI companies a stamp of approval from regulators. The discussions come as some in Silicon Valley have floated a variety of proposals for how the society could adapt to the new economic realities of the AI era. Ideas include the creation of shared income or a sovereign wealth fund. The White House didn't respond to a request for comment.

We are exclusively reporting that the FDA has launched a safety study of the abortion pill Mifapristone, potentially paving the way for the Trump administration to restrict its distribution and use.

That comes after some in the anti-abortion movement had accused the administration of dragging its feet on the review to avoid political controversy before the midterm elections, an allegation that the White House and FDA have denied. Reporter Liz Essley Whyte has more.

Liz Essley Whyte (4:26)

This is a study that Secretary Robert F. Kennedy Jr. and former FDA Commissioner Marty McCary had promised to Republican lawmakers, but sources inside the administration had been saying that it hadn't really gotten started.

Now, we know that it's been kicked into high gear. In addition to having its own study done with its own drug safety monitoring systems, we know that the administration is considering hiring a contractor to look at additional data on the safety of the abortion pill and the way it is currently used.