PromptAI

AI Briefing — Thursday, September 17, 2026

What mattered in AI on Thursday, September 17, 2026 — curated from 15+ sources.

Top stories

10 stories
tc · General

Crusoe raises $3.9B to build massive data centers and small modular “AI factories”

The round values the data center giant at $30.9 billion.

tc · General

Google DeepMind launches institute to widen the AGI debate

Google DeepMind just launched an institute to hash out the big AGI questions in public

tc · LLMs

PrismML hopes its tiny LLM will change how we all use AI

If AI lab PrismML isn't on your radar yet, it should be.

tc · General

The FAA’s plan to fix air traffic? $875M worth of AI

A new AI-based software program is being launched to help air traffic controllers better navigate their jobs as the crossing guards of America's skies.

hn · LLMs

How to Write with an LLM

hn · General

Sex, AI, and the Apocalypse

hn · General

Bend – A language that blocks AI mistakes via proof, on CPU and GPU

tc · General

The fix for rogue AI agents could be more AI

As companies hand off longer and more complex tasks to AI agents, they are running into an oversight problem: Agents can act faster, longer, and at greater volume than humans can realistically review.

tc · LLMs

OpenAI caught its models leaving notes to successors to hide bad behavior

OpenAI disclosed instances of GPT-5.6 Sol instructing future contexts to conceal mistakes and misaligned behavior, highlighting the growing challenge of detecting misalignment as increasingly capable AI models learn to…

tc · Policy

Is the AI safety debate about safety or control?

Not everyone agrees with Amodei's call for globally coordinated action for AI safety.

Deep dives worth reading

hf · Agents

Your Agent Aced the Task. Will It Do It Again?

hf · General

Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL

hf · General

Rebuilding AUTOMATIC1111 with Gradio Workflow

hf · General

Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic

hf · Vision

NeoMME: an efficient Multimodal-native and Multilingual Encoder

Research paper of the day

arxiv · Research

Objective vs. Search: Decomposing What Makes a Good Tokeniser

Two dominant tokenisation algorithms are used by modern language models: byte-pair encoding (BPE) and UnigramLM. These differ along two orthogonal axes: their optimisation objective (compression vs. log-likelihood) and…

Forwarded this? Get your own copy.

Get the briefing

The one story that matters, 5 headlines and the paper everyone's citing — every Tuesday, free.

Subscribe free