Global AI Weekly - Issue 103 - Global AI Community

In this issue

Highlights

3 stories

cnbc.com

OpenAI wins $200 million U.S. defense contract

OpenAI has landed a significant $200 million contract with the U.S. Department of Defense, marking a major step in its collaboration with defense-tech startup Anduril. This partnership highlights OpenAI's growing role in integrating advanced AI technologies into national security efforts. The deal demonstrates the increasing reliance on AI for defense applications and strengthens OpenAI's position in this critical sector.

antitrust-intelligence.com

Italy Launches Probe into AI Firm DeepSeek Over Misinformation Risks

The Italian Competition Authority (AGCM) is investigating Chinese AI startup DeepSeek for potentially misleading consumers by not sufficiently warning about the risks of false or misleading content produced by its AI models. The probe aims to ensure transparency and accountability in how the company informs users about the limitations of its technology. This action reflects growing concerns over ethical practices in the rapidly evolving AI industry.

venturebeat.com

Groq just made Hugging Face way faster — and it’s coming for AWS and Google

In this issue

Research

3 stories

time.com

ChatGPT's Impact On Our Brains According to an MIT Study

A new study from MIT Lab scholars explored how using ChatGPT influences brain activity during SAT essay writing. Researchers compared participants writing essays on their own versus with the AI tool, observing notable differences in mental engagement. The findings shed light on the cognitive effects of AI assistance in academic tasks.

simonwillison.net

Design Patterns for Securing LLM Agents against Prompt Injections

A group of 11 researchers from prominent organizations such as IBM, Invariant Labs, ETH Zurich, Google, and Microsoft have collaborated on a paper addressing the challenges of securing language model-based agents against prompt injection attacks. The work explores practical design patterns and strategies to enhance the security of large language models (LLMs) in various applications. This is a notable contribution to advancing the field of LLM security and ensuring safer implementations.

huggingface.co

The Diffusion Duality

In this issue

Video

2 stories

youtube.com

Humans in AI - Geert van der Cruijsen

In this episode of Humans in AI, Geert van der Cruijsen from the Netherlands shares his perspective on how AI is revolutionizing the world of software development. He explains how these advancements are streamlining processes and helping developers achieve faster results. With engaging insights and a dose of humor, this discussion is perfect for anyone curious about the evolving relationship between technology and human creativity.

youtube.com

Foundry Local & There's no reason not to fine tune anymore

This episode of AI Show LIVE dives into the latest updates in the world of AI. It introduces Foundry Local, highlighting its features and uses, and discusses why fine-tuning AI models is now more accessible and beneficial than ever. Whether you're an AI enthusiast or a developer, this session offers practical insights and exciting advancements to explore.

In this issue

Articles

2 stories

blog.langchain.com

The Hidden Metric That Determines AI Product Success

Some AI products achieve massive adoption while others face challenges in gaining traction. Drawing from a decade of experience building AI tools and observing countless industry launches, this article explores the often-overlooked factors that influence success. It sheds light on a critical yet hidden metric that can make or break an AI product's journey in capturing the market.

marktechpost.com

How Do LLMs Really Reason? A Framework to Separate Logic from Knowledge

This article introduces a dual-metric framework designed to evaluate how large language models reason across different domains. It uses the "Knowledge Index" to measure stored knowledge and "Information Gain" to assess logical reasoning capabilities. This approach aims to provide a clearer understanding of the balance between logic and knowledge in LLMs, helping to pinpoint their strengths and weaknesses.

In this issue

Upcoming Events

1 story

agentcon.dev

June: AgentCon 2025

The AI Agents World Tour, part of AgentCon 2025, is a global series of one-day events tailored for developers working with autonomous AI agents. We are now in full swing bringing together top engineers and researchers. Next week we will share a recap of the AgentCon month of June. It's a unique opportunity to connect, learn, and shape the future of AI technology, so check if there is an AgentCon event near you during the next week, or stay tuned for future events.

In this issue

Code

2 stories

blog.cloudflare.com

Connect any React application to an MCP server in three lines of code

We're excited to announce the open-sourcing of use-mcp, a lightweight React library that makes connecting your React applications to any MCP server incredibly simple—just three lines of code! Alongside this, we’re also releasing our AI Playground, a ready-to-use chat interface designed to seamlessly interact with remote MCP servers. Both tools are designed to streamline development and enhance connectivity with minimal effort.

cloud.google.com

Build and Deploy a Remote MCP Server to Google Cloud Run in Under 10 Minutes

Discover how to set up a remote Model Context Protocol (MCP) server on Google Cloud Run in just minutes. This guide walks you through creating a secure and scalable solution for integrating external context with Large Language Models (LLMs). Perfect for developers seeking efficient deployment and reliable performance.

In this issue

Podcast

1 story

youtube.com

Sam Altman | The Future of AI

This week’s episode features a lively conversation with Sam Altman, CEO of OpenAI, as he shares insights on the transformative possibilities of AI and what lies ahead. From AI’s role in groundbreaking scientific discoveries to the risks and rewards of superintelligence, the discussion also touches on the human need for connection, updates from OpenAI, and recent news involving Meta and Scale AI. Packed with future-focused predictions and candid brotherly banter, it’s an engaging look at the intersection of technology and humanity.