Story summary
Scott Alexander's weekly open thread for Astral Codex Ten invites readers to discuss any topic freely. The ACX survey closes tomorrow, and participants must complete it before then to be included in its results. Additionally, Effective Altruism college meetups have been added at Universidad de Sevilla, Universidade Federal De Santa Catarina, and Carnegie Mellon University; time changes are also noted for other scheduled events. Alexander mentions resuming book reviews that week after technical issues prevented their publication the previous Friday.
Alexander highlights a significant development in AI safety with Sam Altman, Dario Amodei, Demis Hassabis, and Elon Musk agreeing to pace AI advancements responsibly, emphasizing continued work is necessary despite progress. He lists new funding opportunities for startups working on technical AI safety or infrastructure: "pre-seed funding of up to $2m to develop an idea, seed funding of up to $20m to build an organization at scale."
Written for “Open Thread Discussion” on 2026-09-14,
grounded in this article and the 0 other(s) covering the same event.
Post about anything you want, ask random questions, whatever.
asserted
you → post → questions
ACX has an unofficial subreddit, Discord, and bulletin board, and in-person meetups around the world.
asserted
ACX → have → world
Most content is free, some is subscriber only; you can subscribe here.
asserted
you → subscribe → ?
1: ACX survey closes tomorrow, take it now or lose your chance!
2:
asserted
survey → close → chance
3: Sorry for no book review last Friday; I was having technical problems with the site.
asserted
I → have → site
The book review posts will resume this week.
asserted
posts → resume → ?
You’ve probably seen that Sam Altman, Dario Amodei, Demis Hassabis, and Elon Musk agreed to “pace the frontier” of AI, with Dario’s essay here as the starting point for a concrete plan.
asserted
Altman → see → plan
I’m very happy with this development, but it’s one step on a long road, and we can’t rest on our laurels.
asserted
we → ’m → laurels
Lots of things are moving now, there’s a lot of interest in the field, and the opportunities to get involved have never been better.
asserted
opportunities → move → field
Coefficient Giving (a giant effective altruist funder, formerly known as Open Philanthropy) is launching Project Tailwind, a “call for founders to launch ambitious new AI safety initiatives”.
asserted
founders → know → initiatives
If you want to fund a charity/startup working on technical AI safety or the surrounding infrastructure (see a list of specific ideas here), they’re offering “pre-seed funding of up to $2m to develop an idea, seed funding of up to $20m to build an org at scale, and more for exceptional opportunities where teams have a strong track record”.
asserted
teams → want → record
6: Fields Medalist Jacob Tsimerman is founding the Mathematical AI Safety Institute, an effort to get mathematicians involved in AI safety research.
asserted
mathematicians → found → research
They are “aiming to hire 10-30 mathematicians to join us in the Bay Area by January, and scale up to 30-100 for September 2027”.
asserted
They → aim → September
If you’re a mathematician who’s depressed about Navier-Stokes and wondering what’s left for human mathematicians in a world where AI capabilities are growing exponentially, this is one possible answer - it’s novel, important, and un-automatable.
asserted
it → ’re → world
Some frontier AI company employees have expressed an interest in holding their CEOs to the commitments they’ve just made.
asserted
they → express → commitments
Pamela Mishkin has organized unded the Coalition Of Concerned AI Staff to unite employees across major companies.
asserted
Mishkin → organize → companies
And Cassie Pritchard, who I disagreed with about happiness research here but who’s been impressively on-point about AI, is trying to organize a safety-focused labor union; you can reach out to her here.
asserted
you → disagree → her
This is why you need to keep a couple of socialists around sometimes!
asserted
you → need → socialists
8: What if you’re not a startup founder, mathematician, or AI company employee?
asserted
you → ’re → ?
I’m still working on figuring out a good answer to this, but here is a boring boilerplate list of things:
asserted
list → work → things
This is slightly complicated, because PauseAI Global and Pause AI USA recently split acrimoniously.
asserted
Global → split → ?
I take PauseAI Global’s side, but it’s not like the Americans are monsters, just hostile and annoying and extremely eye-rolly.
asserted
Americans → take → side
If you’re non-American, you’ll want the Global org at https://pauseai.info/.
asserted
you → ’re → https://pauseai.info/.
If you’re American, you can either ask the Global org how much progress they’re making towards expanding into the US, join the American org and tolerate their flaws, or join StopTheRace (a smaller group I like in the same space).
asserted
I → ’re → space
Pause protest being organized in DC for 9/19!
asserted
protest → pause → 9/19
Add your name to the assurance contract for MIRI’s march on Washington (they’ll do it if they get 100,000 signatories).
asserted
they → add → signatories
EA has lots of institutional money, and the billionaires have most of the obvious things covered, but some partisan organizations, and especially candidates, can’t take certain types of institutional money and need small donors.
asserted
organizations → have → donors
I don’t have a good list of these right now, so you can do your own research or wait to see if I come up with some later.
asserted
I → have → some
Read this year’s ACX voting guides (once they exist), which will highlight local candidates in your area with good opinions.
asserted
which → read → opinions
If you’re interested in a career in AI safety and have relevant skills, get in touch with 80,000 Hours, who offer a career guide, AI-based career counseling, and (occasionally, although they’re very busy) human career counseling.
asserted
they → ’re → guide
Attend a local ACX/EA meetup group, which will get you in touch with other concerned people and maybe help you come up with ideas or organize something.
asserted
you → attend → something
I’ll keep posting opportunities on Open Threads as they come up.
asserted
they → keep → Threads
