Chain of News Digest

Chain of News 02/04/2026

02/04/2026
**Top Story** A critical supply chain attack targeted the AI ecosystem on March 31, 2026, when malicious actors published compromised versions of LiteLLM—a widely deployed LLM proxy library—to PyPI. The poisoned packages were live for 40 minutes, enabling potential remote code execution on any system that installed them, representing an extreme risk for thousands of production applications relying on the library for model routing and fallbacks. This incident underscores the acute vulnerability of the AI stack's dependency chain, where a single compromised tool can cascade into systemic failures, and should force every team using AI libraries to immediately audit their `requirements.txt` and implement strict package signature verification. **AI Models & Research** Open models have reached a pivotal milestone: GLM-5 and MiniMax M2.7 now match closed frontier models on core agentic tasks like file operations, tool use, and instruction following, but at a fraction of the cost and latency, according to new evals. This threshold crossing fundamentally alters the economics of deploying capable AI agents, making self-hosting and fine-tuning a default consideration rather than a compromise. Google launched Vids, a free, AI-powered video creation and editing tool integrated into Workspace, which could disrupt simple marketing and tutorial video production but is less directly relevant to core developer workflows. Moonlake's research presents an interactive, multiplayer world model approach bootstrapped from game engines, pointing toward more dynamic and efficient simulation environments for agent training, though it remains a research prototype. **Developer Tools & Frameworks** Docker Hub has significantly expanded its role as an AI model repository by adding Gemma 4 as a generally available OCI artifact, making it trivial for developers to pull and run the latest Google model with `docker run`. Docker Desktop's Offload feature is now generally available, a major usability breakthrough that allows running Docker in restrictive VDI or remote environments by offloading the engine to a local helper, finally bringing full Docker capability to millions of locked-down enterprise developers. Codex has introduced pay-as-you-go pricing for its ChatGPT Business and Enterprise tiers, providing teams with a lower-commitment, flexible scaling path to adopt AI coding assistance beyond fixed-seat subscriptions. The Axios npm supply chain attack, where a malicious dependency executed a C2 call within 1.1 seconds of installation, serves as a brutal reminder that even trusted packages can be weaponized, demanding rigorous use of lockfiles, private registries, and software composition analysis tools. **Industry & Business** OpenAI has acquired TBPN, the popular live-streaming interview show focused on tech and AI leaders, in a move to accelerate global conversations and support independent media, effectively bringing a key community platform and its audience in-house. This acquisition is less about technology and more about narrative control and direct engagement with builders and businesses, bypassing traditional media channels. BizNode, a new business intelligence tool, highlights the growing trend of local-first AI by using Ollama with Qwen3.5, promising that all data processing occurs on a user's own hardware for true privacy, a compelling sell for regulated industries. The LiteLLM breach has immediate business implications, likely triggering emergency security reviews and potential liability discussions for any SaaS provider that integrated the library, as the attack window, though short, was sufficient for widespread distribution. **Worth Watching** VelesQL proposes a paradigm shift by enabling semantic, meaning-based search directly within SQL, bridging the gap between traditional relational queries and vector/knowledge graph search without requiring a separate API or client, which could simplify many AI-enhanced application backends. The story of CodeAtlas, a developer tool built by a 16-year-old out of frustration with project navigation, is a noteworthy anecdote about grassroots tool innovation and identifying pain points that established IDEs overlook. A "quiet" April Fools' day in AI news suggests the industry's maturation, where major players focused on substantive releases and security patches instead of pranks, a sign of the stakes now involved. Finally, the recurring theme across several items—from Docker Offload to BizNode—is a clear industry pivot toward solving the practical deployment, security, and privacy hurdles that stand between AI research and reliable, scalable production systems.

Today's Stories

Today's articles

GNews: AI Italia

L’intelligenza artificiale sale a bordo: ChatGPT sbarca su Apple CarPlay con iOS 26.4 - Andrea Galeazzi

L’intelligenza artificiale sale a bordo: ChatGPT sbarca su Apple CarPlay con iOS 26.4 Andrea Galeazzi

02/04/2026
Dev.to JavaScript

I built a Developer Tool at 16 (CodeAtlas)

I built a developer tool at 16, and it didn’t start as anything impressive or serious. It started simply because I was frustrated - which I didn’t even realise at first, as I just assumed it was all just part of programming. Every time I opened a new codebase, especially anything larger than a school project, it felt like I was reading something that was barely understandable for most humans. Files depended on other files, functions called things that were buried deep in completely different fol

02/04/2026
Dev.to AI

BizNode uses Ollama (Qwen3.5) running locally on your hardware — your data never leaves your machine. True AI privacy

The future of business is not about working harder; it is about working smarter with intelligence that operates around the clock. Imagine a team of employees who never sleep, never take holidays, and are driven by pure logic rather than human emotion. This is the reality BizNode creates through its unique fusion of artificial intelligence and autonomous operational nodes. What Is BizNode? BizNode is not just a software tool or a simple chatbot service. It represents a new paradigm in business in

02/04/2026
Dev.to AI

Best Openclaw Alternatives For Secure, Fully Managed Agents (2026 Buyer's Guide)

OpenClaw is the most capable open-source personal AI agent framework available right now. But deploying it in production comes with a real cost: self-hosting means you're managing VPSs, maintaining Docker container orchestration, and debugging OAuth authentication flows. Every week, indefinitely. This guide evaluates the top alternatives across two categories to help you escape that burden: fully managed OpenClaw hosting providers and general personal AI assistants. We wrote this guide for techn

02/04/2026
Latent Space

Moonlake: Causal World Models should be Multimodal, Interactive, and Efficient — with Chris Manning and Fan-yun Sun

We cap out our World Models coverage with one of the most exciting new approaches - long running, multiplayer, interactive world models built with agents bootstrapped from game engines!

02/04/2026
Dev.to AI

What if SQL could search by meaning? Meet VelesQL

You know SQL. You have been writing SELECT, WHERE, and JOIN for years. But the moment you need to search by meaning, traverse a knowledge graph, or rank results by relevance, SQL cannot help you. You reach for a proprietary API, a different client, a different mental model. What if you did not have to? VelesQL is a query language that starts where SQL stops. It keeps the syntax you already know and adds three things SQL never had: vector similarity search ( NEAR ), graph pattern matching ( MATCH

02/04/2026
LangChain Blog

Open Models have crossed a threshold

💡 TL;DR: Open models like GLM-5 and MiniMax M2.7 now match closed frontier models on core agent tasks — file operations, tool use, and instruction following — at a fraction of the cost and latency. Here's what our evals show and how to start using them

02/04/2026
Dev.to DevOps

AI Agent Supply Chain Attacks: What the LiteLLM Breach Means for Your Stack

The morning of March 31, 2026, started badly for the AI ecosystem. A malicious actor had slipped compromised versions of LiteLLM — one of the most widely-deployed LLM proxy libraries in production — onto PyPI. The poisoned packages were live for 40 minutes . That was enough. Mercor, the AI-powered hiring platform backed by top-tier VCs, disclosed it had been hit. And Wiz's cloud scanning data made the scale immediately clear: LiteLLM is present in 36% of cloud environments . ~500,000 machines re

02/04/2026
Dev.to JavaScript

The Axios npm Supply Chain Attack (March 2026): A 2-Second Breach Window That Compromised the JavaScript Ecosystem

You run npm install . It’s muscle memory at this point. Dependencies resolve. Progress bar moves. Nothing unusual. 1.1 seconds later — your machine has already made an outbound call to a command-and-control (C2) server. Not after install. Not when you run the app. During install. Before npm even finishes. That’s exactly what happened in the March 2026 npm supply chain attack involving [email protected] and [email protected] . And if you installed either version, you didn’t just pull a library—you executed

02/04/2026
The Verge AI

OpenAI just bought TBPN

OpenAI has purchased TBPN, the viral online talk show that often interviews AI executives and other tech leaders. The show - which goes live every weekday, often for a three-hour duration - considers channels like Bloomberg and CNBC as its competition and counts OpenAI CEO Sam Altman, as well as executives from Meta, Microsoft, Palantir, […]

02/04/2026
Docker Blog

Gemma 4 is Here: Now Available on Docker Hub

Docker Hub is quickly becoming the home for AI models, serving millions of developers and bringing together a curated lineup that spans lightweight edge models to high-performance LLMs, all packaged as OCI artifacts. Today, we’re excited to welcome Gemma 4, the latest generation of lightweight, state-of-the-art open models. Built on the same technology behind Gemini,...

02/04/2026
Google AI Blog

Create, edit and share videos at no cost in Google Vids

Google Vids logo surrounded by various video editing UI

02/04/2026
Docker Blog

Docker Offload now Generally Available: The Full Power of Docker, for Every Developer, Everywhere.

Docker Desktop is one of the most widely used developer tools in the world, yet for millions of enterprise developers, running it simply hasn’t been an option. The environments they rely on, such as virtual desktop infrastructure (VDI) platforms and managed desktops, often lack the resources or capabilities needed to run Docker Desktop. As enterprises...

02/04/2026
Cloudflare Blog

Why we're rethinking cache for the AI era

The explosion of AI-bot traffic, representing over 10 billion requests per week, has opened up new challenges and opportunities for cache design. We look at some of the ways AI bot traffic differs from humans, how this impacts CDN cache, and some early ideas for how Cloudflare is designing systems to improve both the AI and human experience.

02/04/2026
OpenAI Blog

OpenAI acquires TBPN

OpenAI acquires TBPN to accelerate global conversations around AI and support independent media, expanding dialogue with builders, businesses, and the broader tech community.

02/04/2026
OpenAI Blog

Codex now offers more flexible pricing for teams

Codex now includes pay-as-you-go pricing for ChatGPT Business and Enterprise, providing teams a more flexible option to start and scale adoption.

02/04/2026
Latent Space

[AINews] A quiet April Fools

a quiet day

02/04/2026
Simon Willison

March 2026 sponsors-only newsletter

I just sent the March edition of my sponsors-only monthly newsletter . If you are a sponsor (or if you start a sponsorship now) you can access it here . In this month's newsletter: More agentic engineering patterns Streaming experts with MoE models on a Mac Model releases in March Vibe porting Supply chain attacks against PyPI and NPM Stuff I shipped What I'm using, March 2026 edition And a couple of museums Here's a copy of the February newsletter as a preview of what you'll get. Pay $10/month

02/04/2026
Dev.to LLM

Top 5 AI Agent Security Tools for Developers in 2026

In a shocking turn of events, a single, well-crafted adversarial input was able to bring down an entire AI-powered customer support system, resulting in over $1 million in lost revenue and countless hours of downtime. The Problem import torch from transformers import AutoModelForSeq2SeqLM , AutoTokenizer # Load pre-trained model and tokenizer model = AutoModelForSeq2SeqLM . from_pretrained ( ' t5-base ' ) tokenizer = AutoTokenizer . from_pretrained ( ' t5-base ' ) def generate_response ( user_in

02/04/2026
Dev.to LLM

Hướng Dẫn Tự Tạo Mã Claude

Điểm chính Vụ rò rỉ mã nguồn Claude Code đã tiết lộ một codebase TypeScript gồm 512.000 dòng vào ngày 31 tháng 3 năm 2026. Kiến trúc cốt lõi là một vòng lặp while gọi API Claude, phân phối các lệnh gọi công cụ và trả về kết quả. Bạn có thể tự xây dựng phiên bản của riêng mình với Python, Anthropic SDK và khoảng 200 dòng mã cho vòng lặp chính. Hướng dẫn này sẽ phân tích từng thành phần và chỉ cho bạn cách tái tạo chúng. Hãy dùng thử Apidog ngay hôm nay Giới thiệu Vào ngày 31 tháng 3 năm 2026, Ant

02/04/2026