← All briefings
Tuesday, June 23, 2026

Lumis Daily Briefing — Jun 23, 2026 — Steam Machine is back, and Valve is swinging for the living room again

This is what Lumis subscribers got in their inbox this morning — synthesized from Hacker News, arXiv cs.AI, The Batch, and Latent Space.

Copied!
Top 3 Stories
#1 RELEASE

Steam Machine Returns: Valve Bets Big on Living Room PC Gaming

With 1,390 upvotes and 1,243 comments, this is the day's most-discussed tech event. Valve's revived Steam Machine challenges consoles directly, and if it ships with SteamOS maturity behind it, the living room gaming market faces real disruption.

#2 POLICY

Flock's License Plate AI Used by Police Chiefs to Stalk Women

IPVM documents police chiefs using Flock's mass surveillance network to track women without warrants — a concrete abuse case that will accelerate legislative pressure on automated license plate reader (ALPR) vendors and their law enforcement contracts.

#3 POLICY

Canada Commits to 10 New Nuclear Reactors by 2040

Canada's federal nuclear strategy is the largest Western reactor buildout announced in decades, directly responding to AI data center power demand and energy sovereignty goals. It sets a policy template other G7 nations are watching closely.

More from today
MARKET

Chevron & Microsoft Lock In 20-Year West Texas Power Deal

A two-decade energy agreement between an oil major and the world's second-largest cloud provider signals that hyperscalers are now structuring infrastructure commitments at the scale of industrial utilities — reshaping how data center power is sourced and priced.

BREAKTHROUGH

Moebius: Tiny 0.2B Model Matches 10B Image Inpainting Quality

HUST researchers achieve 10B-class inpainting performance at 0.2B parameters, a 50x efficiency gain that makes production-grade image editing viable on edge devices and dramatically cuts inference costs for creative AI pipelines.

RELEASE

Oak: A Git-Style Version Control System Built for AI Agents

Oak reimagines source control around agentic workflows, where non-human actors need deterministic, auditable commit histories. As agent-driven development scales, purpose-built version control becomes critical infrastructure.

RESEARCH

LLM Reasoning Shows Irrationality Beyond Value Misalignment

New arXiv research finds LLMs exhibit systematic irrationality in reasoning chains that is distinct from — and compounds — known value misalignment issues, complicating alignment strategies that treat reasoning and values as separable problems.

RELEASE

GLM-5.2 Local Deployment Guide Drops via Unsloth

Unsloth's step-by-step guide for running GLM-5.2 locally lowers the barrier for self-hosted deployment of one of the strongest open-weight multilingual models, accelerating adoption in privacy-sensitive and air-gapped enterprise environments.

Get it in your inbox

Get tomorrow's briefing delivered at 07:00 UTC.

Lumis synthesizes the top AI and tech developments into a sharp 3-item briefing — personalized to your sources, delivered before your day starts.

Founding members: $9/month — locked for life
Start 7-day free trial →

7 days free. Cancel anytime. No credit card needed.