Live
OpenAI announces GPT-5 with real-time reasoning capabilities EU AI Act enforcement begins — what it means for US companies Google DeepMind's Gemini Ultra beats GPT-4 on 12 new benchmarks Nvidia posts record $22B quarter fueled entirely by AI chip demand Meta open-sources Llama 3.5 — most capable open-weight model yet Anthropic raises $4B from Amazon in landmark AI safety investment
Vol. III, No. 847 Monday, March 31, 2026 Morning Edition

Curated AI Intelligence

Where AI meets
the world.

Research breakthroughs, industry shifts, regulatory moves — everything that's reshaping intelligence, in one place. No hype, no filler.

40K+
Readers daily
200+
Sources monitored
12K+
Newsletter subscribers

What everyone's reading

EU Parliament building abstract
Policy

EU AI Act Enforcement Begins — The First 90 Days Will Set the Tone

Marcus Webb · 5 min read
Data center chip visualization
Industry

Nvidia's $22B Quarter Shows No Sign of Cooling — And That's the Problem

James Okafor · 4 min read
Open source AI model illustration
Models

Meta's Llama 3.5 Is Here — And It Closes the Gap With Closed Models Faster Than Expected

Priya Nair · 6 min read

Fresh off the wire

AI safety research abstract
Research

Anthropic's Constitutional AI v2 Teaches Models to Disagree With Themselves

A new layer of self-critique during pretraining reduces harmful outputs more than RLHF alone.

Priya Nair · 3 min
Voice waveform visualization
Tools

The Race to Build Real-Time Voice AI That Doesn't Sound Robotic

ElevenLabs, Hume, and OpenAI's Advanced Voice Mode are converging — and quickly.

Alex Rivera · 5 min
Code editor with AI suggestions
Industry

30% of GitHub's New Code Is AI-Written. Developers Are Split on What That Means

Copilot's latest usage report reveals a quiet transformation in how software gets built.

Dana Moore · 7 min
Molecular structure visualization
Research

AlphaFold 3 Just Helped Find a Drug Candidate in 72 Hours. Here's the Full Story

What seemed like science fiction two years ago is now a repeatable workflow at major pharma labs.

Sarah Chen · 9 min
Washington DC Capitol building
Policy

Senate Proposes First Federal AI Liability Framework — Who Gets Blamed When Models Fail?

A bipartisan bill attempts to define accountability in an industry that's moved faster than law.

Marcus Webb · 6 min
Compact neural network diagram
Models

Small Language Models Are Getting Surprisingly Good — And That Changes the Economics of AI

Why Phi-3, Mistral 7B, and Gemma are forcing a rethink of what "good enough" actually means.

James Okafor · 5 min