The Morning Download: Anthropic Wants to Pump the Brakes on AI

Good morning. Citing the "immense implications" of self-improving AI arriving before society is prepared, Anthropic called upon top AI labs to consider slowing the pace of development.

The ability to slow global AI development would “likely be a good thing,” the company said Thursday in a blog post that disclosed internal data documenting how quickly its most advanced models are improving. Highlights from the WSJ story:

The post, written by the head of its internal research institute and a company co-founder, noted that model advances appear to be on a path toward “recursive self-improvement,” when AI systems can improve on their own without human intervention. Some AI insiders have seen that threshold as a potential marker of danger and enormous societal upheaval.

“We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology,” the post, written by Marina Favaro and Jack Clark, says. It proposes a global agreement on how to potentially slow development and a mechanism for verifying that competitors are respecting it.

A slowdown, one can argue, would benefit Anthropic given that it is advancing so quickly on the technological and commercial level. Even so, the most important signal may be that recursive self-improvement “could come sooner than most institutions are prepared for.”

While that outcome isn’t inevitable, company leadership needs to brace for the possibility that AI’s impact could suddenly and sharply accelerate beyond the current level. That might unlock tremendous opportunities, the very pace and intensity of change means it wouldn’t be benign, either.

Shares opened at $68 before easing to $60.38, ending the session with a gain of 0.6%, Barron's reports.

Chief Executive Rajeeb Hazra told WSJLI’s Isabelle Bousquette that he’s not worried too much about day one stock volatility, but then, for leaders of quantum computing companies the ability to take a long term view is uniquely imperative.

“We can never sit down today and say what [quantum] is going to be able to enable in the next 10 years. The best thing we can do is build the technology to the requirements that we see as most valuable,” he said.

The challenge is balancing technical advances with an eye on which of them have commercial relevance. “At the end of the day, is it just another amazing piece of science or does it have a home in a product or service that actually creates customer value for us?”

Global opportunity in AI. Nikisha Alcindor is a professor at the Rochester Institute of Technology’s Saunders College of Business. In her research, she applies machine learning and AI to improve the success rates of mergers and acquisitions. Alcindor is also president and founder of the STEM Educational Institute, which seeks to end generational poverty through free programming in the areas of STEM, financial literacy, and mental health.

On Thursday, the New York-based nonprofit hosted Mind the Gap: Intelligence in Transition, a Tech Week event in Harlem that addressed ways to prepare the workforce for an AI-driven economy. She interviewed Nvidia’s Bianca Pryor, director, data science for media and entertainment, and Nvidia’s Fatima Tambajang, head of AI ecosystem and strategic partnerships, emerging markets.

Alcindor asked them to describe the relationship between infrastructure, talent, and access, when you are bringing AI to market.

“It’s a little bit of a chicken and egg … the question always is, does infrastructure lead to talent creation or does talent lead to infrastructure demand. That question to be honest is still being answered, because we’re seeing different things in different markets,” Tambajang said.

Tambajang said that sectors like healthcare and financial services in emerging markets place a high priority on data sovereignty, requiring localized data storage solutions and physical GPUs located on their own soil, a setup crucial for building trust with local institutions.

Tambajang said Tunisia, Morocco, Nigeria and Egypt and South Africa are leading AI development in Africa. Tunisia and Egypt boast historically deep roots in rigorous STEM and mathematics education, she said. The primary structural challenge is retention, with a large number of Tunisian doctors moving to France and highly educated Nigerian immigrants moving to the United States. That creates a domestic talent gap, she said as local ecosystems struggle to hold onto skilled workers.

Pondering AI’s impact over some pasta. At SoHo’s Charlie Bird, a small group hosted by Haymaker Group founder J.J. Colao took on some big topics: creativity, entrepreneurship, and AI’s impact on society.

“Everyone is terrified of their position in this ecosystem,” said Ross Buhrdorf, co-founder and CEO of ZenBusiness, a software platform that helps new businesses incorporate. In other words, every job and company is vulnerable to AI disruption.

It’s a conversation that’s felt hard to escape these days with Tech Week events happening throughout New York. And not just because, at one point, the door to the private dining room at Charlie Bird came off its hinges and we were briefly, literally trapped.

One bright spot? AI seems to be making it easier than ever for entrepreneurs to start and run their own businesses. I asked Buhrdorf what are the most popular types of companies using ZenBusiness to incorporate these days. His answer? AI companies and marijuana companies.

— Isabelle Bousquette

• Senior U.S. officials have discussed having the federal government take stakes in major AI companies after OpenAI Chief Executive Sam Altman pitched the idea to the administration last year, the WSJ reports. The Trump administration has announced direct investments in at least 10 companies, including the chip maker Intel.

• The FT reports that Anthropic has installed staff within the National Security Agency to help with the use of its powerful AI model Mythos. 

• U.S.-based employers announced 97,006 job cuts in May with technology remaining the leading sector for staff reductions, according to a report from global outplacement and executive coaching firm Challenger, Gray & Christmas. Employers in the tech sector announced 38,242 cuts last month. For the year, the tech sector has announced 123,653 cuts, up 66% through the same period in 2025.

• As the Phoenix area emerges as a major data-center market, the state’s largest utility is proposing a 45% electricity-rate increase for “extra-large energy users,” primarily data centers, and a roughly 14.5% increase for residential customers, a move that has drawn criticism from companies like Microsoft, but also consumer advocates who warn the plan would shift the financial risks of the AI build-out to households already struggling with high summer electricity bills.

• "