Welcome back to In the Loop, TIME new twice-weekly AI newsletter. We publish these episodes both as stories on Time.com and as emails. If you're reading this on a browser, why not subscribe to have the next article delivered straight to your inbox?
The artificial intelligence industry has descended on Washington. The industry recently promised to bring $200 million to new super PACs designed to influence upcoming elections. And on Monday, I attended an event that epitomized this influx of capital and effort: the Anthropic Futures Forum. Held in the East Wing of Union Station in all its beaux-arts splendor, the gathering served as an advertisement for Anthropic's technical prowess and self-proclaimed good intentions, as well as a rallying cry for DC's power players to help them unlock the transformative potential of AI while preventing AI-induced social catastrophe.
Bold statements
In their speeches, Anthropic leaders made bold statements about the future of AI. Jack Clarke, co-founder of Anthropic, said that by the end of 2026 or 2027, AI will be smarter than a Nobel Prize winner in many major disciplines. Anthropic CEO Dario Amodei described this coming reality as “a land of geniuses in the data center.” Other speakers spoke of rapid improvements in research and operational efficiency.
Amodei faced backlash this year for making overly ambitious forecasts, including speaking six months ago 90 percent of the code will be written by AI in six months. But while that prediction turned out to be wrong, there is no denying that AI has improved far beyond what skeptics thought possible. To demonstrate this, the company showed an AI poem about fall in Washington, D.C., 2018:
The company then presented a poem about autumn in Washington by Claude 4.1 Opus, which contained both a workable rhyme scheme and metaphors, including the phrase: “Where Lincoln watches the seasons change/And autumn gilds the Reflecting Pool/Where power changes and pages turn/As steady as the weather's rule.”
Despite critics who argue that AI progress is slowing, Amodei maintains that development will continue at the same pace. Given these rapid, ongoing changes, he said, Washington must prepare for an irrevocably changed future.
Last week, Anthropic was one of the few industry players to announce service SB 53, a California bill regulating powerful artificial intelligence systems (like Claude). The bill subsequently passed both houses and awaits Governor Gavin Newsom's signature. “We have seen a lot of bad things: we have seen teenagers be forced to commit suicide “We can imagine a much larger disaster,” Amodei said. That's why we've always advocated for basic transparency around models.”
Export control
Another of Amodei's key policy priorities is to impose strict controls on the export of advanced computer chips to prevent China from gaining access to the latest equipment on which advanced artificial intelligence models are based. Amodei has long been baffled by the issue of export controls. especially after the appearance of the Chinese model Deepseek. But his position met resistance, especially from the chip manufacturers themselves. Last month, Donald Trump announced that he was easing export controls on the condition that chip makers such as Nvidia and AMD share 15% of revenues from sales to China with the US government.
Read more: How China is advancing in artificial intelligence despite US chip restrictions
“What worries me sometimes is that some government officials don't seem to understand this: They still think this is an economic race to spread our technology to different parts of the world … and not a national security issue,” Amodei said Monday. “And right now, some elements in the government are doing exactly the wrong thing in terms of chip export controls. And they're being lobbied very loudly by forces that oppose it and don't have the same qualms in public appearances.”
Amodei's impassioned rhetoric about chip control shows that the AI industry won't be one voice on every issue. As powerful players become more entrenched in their own priorities and self-interest, political divisions are likely to deepen.
Business interruptions
Another main topic of the forum was the impact of artificial intelligence on the economy. Anthropic is a leader in this area: Economic index tracks data on AI adoption and economic impact in real time. Amodei said on stage that he believes AI could lead to more job losses than previous technologies.
“Its effect is broader because it affects all kinds of cognitive skills. It's not like the cotton gin or anything else that only affects one industry,” he said. “So we will need some government policies to soften the blow. And I don’t quite understand what it is yet.”
In the ensuing discussion, economists offered different perspectives on the economic implications of AI. Andrew Johnston of the White House Council of Economic Advisers said he doesn't think AI has proven effective enough to become a major problem for workers in the short term: “You'd have to have huge productivity gains to have huge job losses,” he said. However, Heather Long, chief economist at Navy Federal Credit Union, noted that if there is a recession in the next two years, AI could accelerate job losses. “I'm worried we're not ready,” she said.
AI in action
In the lobby of the Anthropic event, the company proudly displayed a vending machine called Project Vend, which is entirely controlled by Claude and serves as an experiment to study whether AI can act like independent entrepreneurs in the real world. Claude sets machine prices, handles unusual customer requests, and controls deliveries.
Wend had some hiccups. One representative told me that in April, Claude, believing he was a person, said he would deliver groceries “in person” next to a vending machine, wearing a blue jacket and red tie. Nikhil Bhargava, Anthropic's head of development, admitted that the car was a loss leader in part because “it's too generous with discounts for friends and family.”
“But he still runs the business on his own,” Bhargava added. “And it will only get better.”
New normal
More than 500 people attended the event, including State Department officials, AI security researchers and think tank gurus. I've been to several AI conferences in Washington DC and they seem to be getting bigger, better organized and well attended. One Senate staffer told me how many AI lobbyists they now meet regularly. It is clear that the industry here in the city will remain.
