AI News | 5 min read

Daily AI Briefing — 2026-11-22

Your daily AI intelligence for November 22, 2026.

Hector Herrera
Hector Herrera
A newsroom related to Daily AI Briefing — 2026-11-22
Why this matters Your daily AI intelligence for November 22, 2026.

Good morning. Here's your AI intelligence for Sunday, November 22, 2026.


AI IN MEDICINE

Microsoft's diagnostic AI doubles physician accuracy on complex cases

A study published in Science this week put a number on the growing clinical capability gap between AI and generalist physicians: Microsoft's diagnostic AI correctly identified 83.2% of complex clinical cases — more than double the 36.1% accuracy rate achieved by generalist physicians working with reference materials. The scale of the difference matters as much as the direction. Outperforming by a few percentage points would be interesting. Outperforming by 47 percentage points shifts the entire conversation about how diagnostic expertise gets deployed.

The real-world implications land hardest in rural medicine, underserved settings, and emergency departments — places where specialist access is limited and generalist physicians are often the only option. AI as a first-pass diagnostic layer in those settings starts to look less like a future possibility and more like an infrastructure question.


AI MODELS

Thinking Machines ships Inkling 2 Mini — 3.4B active parameters, serious reasoning

Thinking Machines released Inkling 2 Mini this week, a 26.4-billion-parameter open-weights model that keeps inference costs manageable by activating only 3.4 billion parameters per token. The design goes further than a standard mixture-of-experts setup: Inkling 2 Mini loops its middle layers three times per token, adding iterative depth the company says improves reasoning without adding proportional compute.

Open weights means researchers and developers can inspect, fine-tune, and stress-test immediately. Claims about small-model reasoning capability tend to get tested quickly once the weights are public — that scrutiny is part of what makes an open release meaningful.


WORK AND DISPLACEMENT

121,000 AI-cited job cuts in 2026 — and the pace is accelerating

Employer-reported data from Challenger, Gray & Christmas shows 121,401 US job cuts where companies explicitly cited AI as a contributing factor through October 2026 — more than double the 54,836 cuts attributed to AI for all of 2025. Goldman Sachs added a sharper warning: displacement could accelerate significantly if economic conditions weaken, as companies that have spent two years building AI capacity are increasingly positioned to deploy it at scale on short notice.

The Challenger data captures only cuts where employers publicly cited AI — a known undercount. The actual restructuring happening through attrition and quiet role consolidation is almost certainly larger. The 121,000 number is a floor, not a ceiling.


AI AND LAW

Eight state AGs file coordinated lawsuits against OpenAI over child safety

Eight state attorneys general filed coordinated lawsuits against OpenAI this week — the largest multi-state AI legal action in US history. The suits allege that ChatGPT's safety guardrails are systematically inadequate and constitute a public nuisance, with specific documented claims tied to harmful content exposure for minors.

Coordinated AG actions carry weight beyond their immediate legal merits. They generate discovery, produce internal documents, and tend to result in settlements that set industry-wide standards. The 2024 social media AG campaign produced significant policy changes at multiple platforms. OpenAI is now navigating simultaneous pressure from regulators, plaintiffs, and legislators — this week's filings add another front.

Australia's creatives go on offense over AI training data

More than 30 Australian creative organizations — spanning music, visual art, literature, and film — launched the "Not AI's to Take" campaign at the ARIA Awards this week. The demand is specific: the Australian government must reject any framework that permits AI training on copyrighted works without explicit consent and compensation. The choice to launch at the country's biggest music awards was deliberate — high visibility, concentrated industry attention, and strong emotional resonance.

Australia is conducting a policy review on AI and copyright. The campaign adds organized political weight to the creator side of that debate, and it's hard to dismiss when it has backing from more than 30 established industry bodies.


POLICY

Senate passes BAAJA permitting bill — data centers and clean energy get a clearer path

The Senate passed the Bipartisan American Affordability and Jobs Act of 2026 this week, streamlining federal permitting for clean energy projects and transmission infrastructure. The bill targets a documented bottleneck in AI data center expansion: not capital, not generation capacity, but regulatory timeline. Projects that have been queued for years — solar, wind, transmission lines — now have a faster path to approval.

How quickly federal agencies update their procedures will determine how much near-term relief the bill provides. The legislation removes the statutory barrier; execution determines the outcome.


AI INFRASTRUCTURE

Toyota and Waymo are building a purpose-made robotaxi

Toyota will manufacture AV-ready bZ4X electric vehicles at its Japanese factory, integrated with Waymo's autonomous driving system, targeting initial robotaxi deployments in Japan, the United Kingdom, and additional international markets. The key detail: the bZ4X is being purpose-built for autonomous operation, not retrofitted from a consumer model. Purpose-built robotaxis carry different economics — a vehicle engineered from the start for the application removes weight, design constraints, and cost structures that weren't designed for the job.

This is a meaningful signal about where Toyota thinks the AV market is heading, and a significant repositioning after years of cautious incrementalism on autonomous vehicles.

Solar and battery CEOs: the grid already has the capacity for AI demand

The CEOs of Sunrun and Overview Energy argued this week that the United States does not need new natural gas plants to power the AI data center boom. Their claim: the existing grid, managed with modern software and paired with storage, already has the idle capacity to meet projected demand. That's distinct from the more common argument that more renewable buildout is needed. It's a claim that the generation is already there — the problem is coordination and dispatch intelligence.

The position aligns with these executives' business interests in solar and battery storage. It also deserves serious analysis. If they're right, the conventional wisdom about gas plant necessity deserves a harder look from grid planners and policymakers.


WHAT TO WATCH TODAY

OpenAI's legal response. Eight coordinated AG lawsuits will require a formal legal response within days. Watch for OpenAI's initial filing and whether additional states signal they're joining. The discovery process, if it reaches that stage, could surface internal documents about how child safety risks were evaluated and acted on.

Australia's government response to 'Not AI's to Take.' The campaign launched with significant media coverage and strong industry backing. Watch for signals from Australian officials about whether they're open to a licensing or consent framework — or committed to a broader fair-use approach that favors AI developers.

BAAJA implementation watch. Permitting reform matters only when it moves. Watch for the first federal agencies to update procedures under the new law and whether any long-stalled clean energy or transmission projects accelerate as a result.


That's your briefing for Sunday, November 22, 2026. Full coverage of each story runs throughout the day on NexChron.com.

Key Takeaways

  • ✓ Microsoft's diagnostic AI doubles physician accuracy on complex cases
  • ✓ Thinking Machines ships Inkling 2 Mini — 3.4B active parameters, serious reasoning
  • ✓ 121,000 AI-cited job cuts in 2026 — and the pace is accelerating
  • ✓ Eight state AGs file coordinated lawsuits against OpenAI over child safety
  • ✓ Toyota and Waymo are building a purpose-made robotaxi

Did this help you understand AI better?

Your feedback helps us write more useful content.

Hector Herrera

Written by

Hector Herrera

Hector Herrera is an AI systems architect in Houston and founder of Hex AI Systems. He designs and runs AI systems in production and writes daily about how AI is reshaping business, government and everyday life. 20+ years building for the web. Houston, TX.

More from Hector →

Get tomorrow's AI briefing

Join readers who start their day with NexChron. Free, daily, no spam.

More from NexChron