Events Email Briefings
rotating globe
Privacy© 2026 Semafor Inc.
  • D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
Events Email Briefings

Intelligence for the New World Economy

  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG
rotating globe
Intelligence for the New World Economy
Semafor World Economy
Reed Albergotti

Reed Albergotti

Tech Editor, Semafor

Email Reed@reedalbergotti

Reed Albergotti is the technology editor at Semafor

It’s time to learn to live with open-source AI

Today’s open-source models from China don’t appear capable of doing much damage yet, but it’s best to prepare.
People walk past the Moonshot booth promoting its Kimi K3 AI model
Go Nakamura/Reuters

AI is just software

Large language models — especially those trained on other labs’ data — are, in the end, just massive, billion-dollar pieces of software.
People walk past the Moonshot booth promoting its Kimi K3 AI model.
Go Nakamura/Reuters

PsiQuantum gets DARPA backing

The $125 million cash infusion adds a new vote of confidence for the company’s efforts to build the world’s first large-scale quantum computer.
A view of a PsiQuantum Wafer, a silicon wafer containing thousands of quantum devices
PsiQuantum/Handout via Reuters

Applied Intuition wants to turn robotics into child’s play

Startup’s new Dana service allows developers to build and test physical intelligence using natural language.
Qasar Younis (CEO) and Peter Ludwig (CTO), Applied Intuition.
Applied Intuition

Kimi K3 threatens AI business models

There’s now an established pattern: American frontier labs come out with state-of-the-art AI models. Chinese firms then allegedly “distill” those models and use them to train new open-source models.
People walk past the Moonshot booth promoting its Kimi K3 AI model
Go Nakamura/Reuters

AI teaches a bitter biology lesson

Human knowledge is getting in the way of scientific progress.
A person works in a science lab
Annegret Hilse/Reuters

Washington confronts China’s open-source models

American AI labs view Chinese open-source models as essentially stolen goods, built with the help of American frontier models to generate synthetic training data — a process known as distillation.

DeepMind’s Hassabis emerges as unifying voice for the AI industry

On Tuesday, Google’s DeepMind co-founder and CEO Demis Hassabis stepped forth as a generally unassailable messenger with an article calling for a self-regulatory body of AI experts.

Apple’s OpenAI lawsuit highlights broader tensions

The case is a gigantic reminder that Apple is under major pressure from AI.

A new plan emerges for AI apocalypse avoidance

The AI Futures Project is advocating for the US and China to cooperate on a temporary pause of frontier AI research so the industry can build some scaffolding around safe development.

Musk acts fast, but can it last?

Cursor and SpaceXAI announced that they collaboratively released a new model, Grok 4.5, designed to work within Cursor.

AI makes mistakes, too

A Codex glitch sent my agent into a loop where it kept reviewing a bunch of data over and over again, racking up tokens — an entirely avoidable problem.

US releases powerful Anthropic model Mythos to some US companies

The move comes the same day as a new OpenAI model sees a limited release.

Hollywood shouldn’t fear AI

AI can help more artists make more movies, rather than replace artists or the art.

OpenAI pulls ahead on custom chips

OpenAI CEO Sam Altman saw from the very beginning that compute infrastructure would become an important battlefront, and his company is getting ahead in a key element that can make models do more, faster and more cheaply.

Demis Hassabis on the link between AI for art and AI for science

For AI to truly revolutionize science, it needs a form of creativity it currently doesn’t possess.

DeepMind Chief Demis Hassabis says Google’s still winning AI talent

Despite the departure of two top AI leaders that sent the company’s stock price down, Hassabis says Google is still on top.

Research pulls back curtain on Claude

“Only about 1.6% of Claude Code’s codebase constitutes AI decision logic, with the remaining 98.4% being operational infrastructure,” a research paper from April explains.