In a time of AI chaos, follow these experts

Share: Link

Aaron Gertler
September 25, 2026

At 80,000 Hours, we spend our days monitoring the AI situation, so we know there’s too much situation these days.

This week’s newsletter is a group effort: we surveyed our colleagues, sifted through our own feeds, and made a list of our favourite follows in the infinite sea of AI news.

The list isn’t comprehensive. We set a limit of 30 sources and focused mostly on those that post often, whether about current events or cutting-edge research.

Blogs (general)
Each of these covers a wide spectrum. Subscribe to any two, and you’ll see the stories that matter most.
  • Don’t Worry About the Vase — Zvi Mowshowitz is a machine. He puts out daily commentary on news stories, new models, and even notable podcast episodes. Not for the faint of heart — there’s a lot of material, with limited context — but if I were trapped on a desert island with only one AI blog, I’d make it Zvi’s.
  • Humans on AI — Biweekly news roundups with brief, well-calibrated comments. If Zvi is too much even for a desert island, this is a great alternative.
  • Import AI — Anthropic cofounder Jack Clark has written 473 weekly newsletters, and he keeps getting better. Thoughtful takes on news and research from the ultimate insider, and a bit of fiction as a treat.
  • Transformer — Written by actual journalists! That means original reporting and more consideration for low-context readers, but also lower information density.
  • AGI.FYI — Not a blog, but a reading list that regularly posts new and classic material. Great for catching up if you’ve only recently started exploring AI safety.
Blogs (specialist)
These are narrower than the previous list, focusing on particular topics and points of view.
  • AI as Normal Technology — Two Princeton academics who think that many claims about superintelligence are incoherent. Many AI safety people cite them as among the best critics of that space.
  • AI Safety in China — 80,000 Hours advisor Sudhanshu Kasewa recommends this for great coverage of a neglected subject.
  • Ajeya Cotra — A great blogger twice over. Planned Obsolescence covers her AI forecasts and a range of technical topics. Good Bones tends to cover the interpersonal aspects of AI safety research, including topics like the corrupting nature of power and what it’s like to work on impossibly hard problems. My fellow writer Avital Morris recommends digging through the archives.
  • Anton Leicht — Hard-nosed, realist writings on AI governance, with an emphasis on the views of policymakers and the plight of middle powers.
  • Dean Ball — Sharp writing on AI governance and possible futures from a former official who served in the Trump administration and helped draft its AI Action Plan. He’s become much more publicly concerned about AI risk in the last year or two, which has made for a lot of interesting posts; I expect more of those now that he’s at OpenAI.
  • Sarah Hastings-Woodhouse — Rambling personal essays with titles like “What the hell just happened?” and “Is everything weird yet?” Fun, accessible, and very human; a writer to commiserate with about the silly and scary parts of our strange era.
  • Understanding AI — Timothy B. Lee is an AGI sceptic who doesn’t expect economic transformation in the next few years. This makes him an interesting contrast to most of the other sources here. Sudhanshu also recommends AI Summer, Lee’s podcast with Dean Ball: “If you buy AGI risk, it’s a good exercise to figure out where your views diverge from Timothy’s.”
Some organisations have great blogs too, including Epoch, METR, Redwood Research, and the AI Futures Project.

Videos
For when you’d rather watch than read, or want something easy to share:
  • Our own channel: AI in Context — Lavishly produced mini-documentaries on the biggest AI stories of the last few years. We aim for compelling, watchable storytelling that you could send to your sharpest sceptical friends. 
  • Rob Miles — The OG. When he launched his channel in 2017, the only AIs most people could name were Deep Blue and the Terminator. But he hasn’t lost a step; his videos still serve as excellent explanations of key concepts.
Podcasts
Content for your walk to work, or your next marathon (if it’s our podcast).
  • The 80,000 Hours Podcast — Hello, again! Smart guests, insanely long interviews: what’s not to like?
  • The Cognitive Revolution — A firehose of information on both safety and capabilities. Unlike most of our recommendations, this podcast covers the way non-AI companies actually use the technology, which provides a different kind of insight than just watching eval scores go up.
  • The Dwarkesh Podcast — Dwarkesh Patel interviews some of the biggest names in AI, and asks some of the best questions you’ll hear on any podcast. I’ve had the privilege of seeing one of his pre-interview Google Docs, and it turns out that some people just outwork everyone else.
Twitter
Nearly everyone we’ve named already is on Twitter: look them up! We’ll also add:
  • Andrew Curran — Conor Barnes, the mastermind behind our job board, calls Curran “oddly sharp and fast for news.” I agree: he’s often the first place I hear about the day’s top story.
  • Daniel Kokotajlo — An OpenAI whistleblower and AI 2027 coauthor who provides bold commentary and forecasts.
  • Gavin Leech — A member of the Humans on AI squad, Gavin is prolific and philosophically inclined, sharing literary references and classy memes alongside research posts.
  • Helen Toner — Think tank director, China specialist, and former OpenAI board member, which adds up to an informed perspective on just about everything she touches.
  • Janus (Repligate) — Very different from the rest of this list. Janus explores how models behave in unusual settings to better understand what’s happening inside, treating them less as tools than as minds worthy of respect. Recommended by external advisor Peter Hartree.
  • Peter Wildeford — A superforecaster and policy analyst whose interests include the way politicians are responding to the current moment.
  • Rob Bensinger — Dispatches from MIRI’s corner of the safety space. Rob argues for global restrictions on AI capabilities research and shares a lot of memes.
  • Rob Wiblin — Our third and favourite Rob. Informed, irreverent takes on AI news and discourse from the original host of a pretty good podcast. Speaking of which:
Featured release: new video podcast

 

Max Nadeau discusses why Coefficient Giving is recruiting founders for a new wave of AI safety nonprofits.Watch the episode →
What we’ve been reading (and watching): Opportunities to get involved:
  • The BlueDot Impact Biosecurity Course helps professionals understand efforts to prevent, detect, and respond to pandemic threats, and identify where they can contribute — no biology background required.
    • Apply by September 27.
  • The Legal Advocates for Safe Science and Technology is accepting applications for its Summer Fellowship. This is a paid, 10-week programme for law students, recent graduates, and practicing attorneys to conduct legal research on AI and biotechnology risk.
    • Apply by October 12.
  • March Against the Machines is a PauseAI UK rally calling on the UK government to introduce frontier AI legislation, following the OpenAI–Hugging Face incident.
    • Taking place December 5 in London.
Have a great weekend,

Aaron
80,000 Hours is a nonprofit that helps people use their careers to solve the world’s most pressing problems. Read our career guide, subscribe to our podcast, speak to our AI advisor, apply to speak with our one-on-one-team, find a job on our job board, and find ways to meet others working to have an impact.

You are receiving this newsletter because you signed up at 80000hours.org or a student activities fair. 80,000 Hours Limited is a not-for-profit company limited by guarantee registered in England and Wales (with registered company number 15746854). 80,000 Hours Third Floor, 20 Old Bailey London, EC4M 7AN United Kingdom

If you only want some of our emails, you can unsubscribe from either our 'job board updates' or 'research updates.' Click here to be emailed a link to edit your preferences. If this link doesn't work, you can email us at team@80000hours.org and we'll update your preferences manually.

Alternatively you can unsubscribe from all our emails and never hear from us again. We hope you won't, though!

Add us to your address book