Anthropic Launches $200M "Economic Futures Research Fund" — Getting Serious About AI's Economic Impact
On July 22, Anthropic published the research agenda for the **Economic Futures Research Fund**, which will invest **$200 million (approximately ¥31 billion)** in studying AI's impact on the economy.
The fund aims to build empirical evidence from external researchers on how AI's rapid adoption will affect labor markets and the broader economy. The five priority research areas are:
Investigating AI's impact on workers at the firm and workplace level
Skill support for navigating AI-driven change
Modernizing income support in preparation for AI-driven job loss
Building mechanisms that let workers share in the gains from AI growth
Generating new evidence on the effectiveness of public investments
Anthropic's "Economic Policy Framework," announced in June, lays out policy recommendations for multiple scenarios, but a key challenge remains the lack of empirical data on which interventions actually work.
Alongside the fund, Anthropic also launched the **Anthropic Economic Index connector**, which lets users search real-world AI usage data directly within Claude. Users can now ask Claude questions like "which occupations use AI the most" or "how are people in Colorado using Claude." Note that the data reflects Claude usage patterns and does not represent the labor market as a whole.
Atoms, the Robotics Startup from Former Uber CEO Kalanick, Raises $1.7B
**Atoms**, the robotics startup founded by Travis Kalanick, has raised **$1.7 billion (approximately ¥263.5 billion)** in a round led by a16z (Andreessen Horowitz). Uber also participated as an investor.
Atoms' stated vision is to "modernize the world with industrial AI," but details about its specific products and technology have yet to be revealed. Since leaving Uber — and passing through the food delivery business — Kalanick has moved into robotics, drawing significant attention.
Massive investments in robotics keep coming, suggesting a trend of AI expanding beyond software into the automation of the physical world.
Monday.com Lays Off 20% of Staff (About 630 Employees) to Focus on AI
**Monday.com**, the project management tool provider, announced on July 22 that it will cut its headcount by **20% (about 630 employees)**. The company cited a shift toward "a leaner, more focused operating model" and explicitly said it will double down on its AI Work Platform.
Layoffs tied to AI-driven strategic pivots are piling up across the tech industry. Monday.com's decision highlights the reality that incumbent SaaS companies are being forced to fundamentally restructure their organizations to compete with AI-native rivals.
Reddit Considers Blocking Google's AI Access — Stock Price Falls
Reddit's stock fell following reports that the company is weighing options to restrict Google's AI access to its content.
Reddit has a data licensing deal with Google, but it appears to be seeking greater negotiating leverage over how AI companies use its content. How content platforms secure compensation from AI companies for data access has become a critical issue for the industry as a whole.
China's Open AI Models Challenge Silicon Valley's Strategy
According to a WIRED analysis (July 22), as access to frontier models from Anthropic and OpenAI becomes restricted, Chinese AI labs are pitching their open-source models as "stable, accessible, and increasingly capable."
Chinese open-weight models are winning over developers worldwide on the strength of their cost performance and freedom of use. As a counterweight to Silicon Valley's closed strategy, China's open-source approach is steadily expanding its influence.
Microsoft Releases Fara1.5-27B, an AI Model Specialized in Browser Operation
Microsoft Research AI Frontiers has released **Fara1.5-27B**, a multimodal AI model that autonomously operates web browsers, on Hugging Face.
The model perceives the browser through screenshots and completes tasks end-to-end by issuing structured tool calls for clicking, typing, scrolling, searching, and more. Notably, it makes decisions based on visual information alone, without relying on the DOM or accessibility tree. It is fine-tuned from Qwen3.5-27B and is publicly available on Hugging Face.
Open models in the computer use space are multiplying, democratizing browser automation — though security concerns are likely to be raised as well.
---