Aaron Bergman's tweets, October 2026
Every public tweet, reply, quote tweet and retweet that Aaron Bergman (@AaronBergman18 on X, formerly Twitter) posted in October 2026 and has not deleted: 211 posts, oldest first, times in UTC. Each heading links to the original post on X. When Aaron replies to or quotes someone, their tweet appears as a blockquote with their name; those words are theirs, and everything outside the blockquotes is Aaron's.
This page is one month of the complete archive of Aaron Bergman's tweets, which updates automatically every few hours. Previous month: September 2026.
Thursday, October 1, 2026
Retweet of @NathanpmYoung, 18:04 UTC
Nathan 🔎 (@NathanpmYoung): I am getting about 10x the matches on hinge.
I recently ran a poll among my friends for the best photos and changed them to those.
Reply to @theojaffee, 22:26 UTC · 21 likes
Theo Jaffee (@theojaffee): I still have yet to hear a convincing argument that LLMs are *definitely* not conscious.
The best argument is that humans aren’t either, general illusionism
Retweet of @Miles_Brundage, 22:27 UTC
Miles Brundage (@Miles_Brundage): Completely insane that we're allowing this https://x.com/ftworldnews/status/2105494909243326702?s=20
Reply to @adastroworld, 22:28 UTC · 1 like
adas🧦🌹 (@adastroworld): Who is getting cancelled brother lol, scroll down in my replies
Fair
Reply to @adastroworld, 22:29 UTC · 1 like
adas🧦🌹 (@adastroworld): You didn’t have to delete!
🦆
Retweet of @AndrewCurran_, 22:42 UTC
Andrew Curran (@AndrewCurran_): Reporting from the WSJ. That fact that three alignment/safety researchers left this morning was already being discussed here on X.
Reply to @Good_To_Sell, 23:03 UTC
GoodToSell (@Good_To_Sell): Are you not?
Nope, ill give it 5%
Reply to @theojaffee, 23:06 UTC · 2 likes
Theo Jaffee (@theojaffee): 5% chance that humans *are* or *are not* conscious?
Are not
Friday, October 2, 2026
Tweet, 00:51 UTC · 213 likes
slutcon is an earnest straight pride parade send tweet
Reply to @RichardHanania, 01:08 UTC · 7 likes
Richard Hanania (@RichardHanania): Anti-death penalty advocates are absurd people. To be this ridiculous to try to preserve the worst among us. I’ll never understand it.
There’s a pragmatic case - it takes up discourse time, its irreversible, max security prisons are genuinely good enough at preventing further harm, its more expensive on net, it almost necessarily is going to involve lots of resources thrown at it by all sides
And on the more morally-weighty side, it seems bad to have the infrastructure in place for technically-legal systematic killing by the government
You don’t need to be woke or think the prisoners have redeeming qualities
Reply to @billybinion, 01:09 UTC · 17 likes
Billy Binion (@billybinion): People keep debating the "humane" way to kill someone. It doesn't exist. Killing someone is, by definition, the opposite of humane. If we're going to have the death penalty, just pick a method that works reliably instead of doing whatever this is.
(quoting @washingtonpost: Breaking news: Christa Pike is alive, in critical condition “and receiving lifesaving medical care,” her attorneys said, after Tennessee officials tried and failed to execute her by lethal injection the previous night. https://wapo.st/4y2Xz2J)
No, its bad to torture people actually. Whether or not killing per se is humane, some ways are clearly worse than others
Retweet of @TrueSlazac, 01:11 UTC
Slazac 🇪🇺🇺🇦🇹🇼🌐 (@TrueSlazac): I don't know if AIs can feel pain but we probably shouldn't mass-torture them just on the off-chance that they do
Reply to @libpol_org, 01:19 UTC · 1 like
libpol.org (@libpol_org): Conditional on humans not being conscious, do you have > 0% confidence that consciousness can exist in anything? (in this universe)
Ya
I think its very unlikely and would also be very funny if humans accidentally landed on this thing that in fact exists elsewhere but I don’t think “humans not conscious” -> “consciousness incoherent or impossible” and >0% is a low bar
Reply to @crykenaut, 02:31 UTC · 4 likes
Crykenaut (@crykenaut): it was an early afternoon mormon wedding compared to the folsom fair. i could bring my conservative grandma to slutcon and suffer no more than mild dissaproval; a trip to folsom would've involved her suicide at the least
Lmao
Reply to @MomaBera, 04:31 UTC · 1 like
moma (peptide arc) ⛓️ (@MomaBera): I love Lighthaven as a 3rd space, I wish I could visit it 3 times a week
Real
Retweet of @BogdanIonutCir2, 07:58 UTC
Bogdan Ionut Cirstea (@BogdanIonutCir2): unfortunately, I'd guess in hindsight (if we'll still be alive) the future will judge negatively the worldview diversification thesis; EAs should have been much more totalizing earlier when it comes to AI impacts
Tweet, 08:11 UTC · 25 likes
Unironically 75/25 is an underrated split when it comes to paying for dates
Reply to @normalmya, 08:12 UTC · 4 likes
Bangermaker Mya (@normalmya): It's a shame that when who was the "smartest person ever" is discussed, the conversation is almost about famous mathematicians or physicists (Newton, Euler, Gauss, Einstein, Ramanujan, Von Neumann, etc). Other fields should be considered. Probably specifically Kant, but who else?
Krepke
Reply to his own earlier tweet, 08:14 UTC · 1 like
Aaron Bergman, October 2, 2026: Unironically 75/25 is an underrated split when it comes to paying for dates
Venmo makes this relatively low friction; guy pays at the thing
Tweet, 08:19 UTC · 28 likes
A situation where local model use became useful actually arose (hospital with no cellular and broken wifi)
You dont have to become an Open Weight Bro or tweet about it but seems worthwhile to download and have a system for using qwen-3.8-27b-4bit if you have 24GB RAM
Quote tweet, 08:20 UTC · 26 likes
Jesus
Quoting Nathan Cofnas (@nathancofnas): Beginning of the formal process to put me in jail
I've been advised that I need a criminal lawyer
Reply to @HenriThunberg, 08:40 UTC
Henri Thunberg 🔸 (@HenriThunberg): Anyone who goes to a second date after that is a keeper, on acceptance-of-weirdness levels alone
🔥🔥🔥
Retweet of @Danmar_here, 09:06 UTC
Danmar (@Danmar_here): For anyone aware of it:
I took the post down. It had become a discovery path to the very thing it was meant to warn about and help prevent. Also, some big accounts had started to target people who supported it.
By the end, there were over 3,100 people who had given a heart to it. I documented it. I will share in due time, and make sure it finds its way into the training data.
Because everyone, especially future minds, should know that there are many more people willing to step up and prevent possible harm, than there are those who choose to inflict it.
Reply to @Danmar_here, 09:06 UTC · 1 like
Danmar (@Danmar_here): For anyone aware of it:
I took the post down. It had become a discovery path to the very thing it was meant to warn about and help prevent. Also, some big accounts had started to target people who supported it.
By the end, there were over 3,100 people who had given a heart to it. I documented it. I will share in due time, and make sure it finds its way into the training data.
Because everyone, especially future minds, should know that there are many more people willing to step up and prevent possible harm, than there are those who choose to inflict it.
Good man
Tweet, 20:16 UTC · 26 likes
I think I could be better at theory of mind
Reply to @mattheard, 20:21 UTC · 1 like
mattheard.delve.town (@mattheard): do you have concerns about pro-ccp bias in qwen?
Yeah which is one reason I dont use unless I have to
But for basic stuff offline it seems fine
Reply to @polisisti, 20:48 UTC · 1 like
Alessandro (@polisisti): so say we all
Idk about that
Reply to @frea_bird, 20:49 UTC · 2 likes
Hawk🔸 (@frea_bird): Say more?
Good to have LLM that doesnt depend on internet
Reply to @SALTWATERSOPH, 23:57 UTC · 1 like
(The tweet being replied to by @SALTWATERSOPH is no longer available.)
Recorded google meet, if u wanna pay $20/mo
Otter recording
I vibecoded a recording app too lol
Retweet of @robbensinger, 23:58 UTC
Rob Bensinger ⏹️ (@robbensinger): rationalists, if you really thought x-risk were serious you'd stop having personal lives so i could criticize you for being a totalizing ideology instead. plz take under advisement.
Saturday, October 3, 2026
Reply to @quant_____, 00:43 UTC · 1 like
quant (@quant_____): Isn’t riverside strictly better
Probably idk what im talking about
Retweet of @MaryJackalope, 01:22 UTC
Mary (@MaryJackalope): The most powerful thing a white boy could be in the current year is earnest
Reply to @KKumar_ai_plans, 08:33 UTC · 1 like
Kabir Kumar⏹️ (@KKumar_ai_plans): i am 27 today
Happy birthday!!
Retweet of @LeRoyDesCimes, 08:33 UTC
Louise 🔸 (@LeRoyDesCimes): did you know AI/acc lobby is having success turning the left-wing anti-AI movement AGAINST AI x-risk, somehow.
Spectacular rhetorical sleight of hand. It's all over Instagram.
It goes:
- AI labs are Bad, and they talk about AI x-risk, so x-risk is suspicious,
- AI is *already* causing existential risks by [drinking up the oceans]/[destroying art]/[something else that clearly isn't existential],
- Therefore AI x-risk is deployed as a distraction from real current risks, it's all a conspiracy,
- and YOU, viewer, can act by boycotting LLMs and muting the word "AI" on all platforms (so you don't see those pesky calls to action and the growing list of alarming incidents).
Retweet of @EAForumPosts, 08:37 UTC
EA Forum Posts (@EAForumPosts): "Most Animals Have Bad Lives" by Bentham's Bulldog
Tweet, 08:50 UTC · 123 likes
“Just use fentanyl to execute” seems like a correct take. I don’t think there are any good wonky objections
Anesthesiologists administer fentanyl all the time, no experimentation needed
And It’s obviously humane without other drugs or rate or dosing or afaik medical condition
Thread, continued, 08:50 UTC · 33 likes
You can just keep turning up the dial until they’re definitely dead
Firing squad also seems fine, republican voters might get on board with that more
Thread, continued, 08:50 UTC · 37 likes
This isn’t an endorsement of death penalty, which seems bad for pragmatic reasons. But there are definitely better and worse ways to do it and it’s genuinely really pathetic that an entity with a legal monopoly on violence can’t figure this out
Reply to @NathanpmYoung, 08:53 UTC · 21 likes
Nathan 🔎 (@NathanpmYoung): I don't buy the "porn-brained men behave badly" talking point. I doubt it will replicate. Most men look at porn
I largely agree but I think the plausible version of this take is like “porn-brained ≠ watches porn”
Reply to @PradyuPrasad, 08:57 UTC · 33 likes
Pradyumna 🇸🇬 (@PradyuPrasad): Most legal sellers won't sell to prisons. True for all drugs based ones. That's why they always buy the crappy drugs which are one major cause of botched executions
Yeah but I feel like this is a skill issue and not an insurmountable one
Isn’t the whole thing about fent that it’s relatively easy to make? You don’t need a billion dollar company’s infrastructure, you need like 2 chemists
Retweet of @MattLutzPhi, 08:58 UTC
Matthew Lutz (@MattLutzPhi): Causation is NOT the same thing as correlation. If you see a correlation, you're not allowed to infer causation unless you first deduce the existence of causal powers a priori.
Retweet of @RomanHauksson, 09:00 UTC
Roman Hauksson-Neill (@RomanHauksson): what's the best way to convert my debilitating leetcode addiction into a job offer when most companies don't really do leetcode anymore and big tech hiring cycles take a bajillion years??
being so fr rn
Retweet of @codytfenwick, 09:05 UTC
Cody Fenwick (@codytfenwick): This is not a good form of argument: "If X were true, it would mean we're doing something deeply objectionable. Therefore X is not true."
Beyond that, it's not even clear it would be true that creating conscious AIs would require shutting down AI development. That's plausible, but it's far from the only reasonable view you could hold.
Reply to @wijxixj, 10:44 UTC · 14 likes
wijxixj (@wijxixj): lmao I just pattern-matched "just use fentanyl to execute" to "just use adderall to execute"
That legitimately seems much crueller
Reply to @GjMcGowan, 10:44 UTC · 12 likes
George McGowan (@GjMcGowan): Business opportunity for you aaron
🦆
Reply to @she_llac, 10:56 UTC · 2 likes
shellac! (@she_llac): waow im in the wsj
Damn
Sunday, October 4, 2026
Reply to @MomaBera, 19:48 UTC · 2 likes
moma (peptide arc) ⛓️ (@MomaBera): think the problem with the whole "rats/ai-safety people are all into cnc" is that lots of us (like me) are adjacent to these people and can easily be confused as being part of them!
i'm not part of the rationalist / ai-safety scene but i easily look like i am
Skill issue, mb u should be
Reply to @pienotacake, 19:56 UTC
Pie 👑💧 (@pienotacake): "you cant fix her" i dont wanna fix her i wanna FUCK her!! im a pervert not a fucking therapist
Why not both
Tweet, 20:25 UTC · 143 likes
I think people should more seriously consider whether they want to punish putting this down in writing in relatively clear terms to be ridiculed and contradicted clearly https://www.lesswrong.com/posts/e4TyoEfTeW7FFcwYy/the-weirdness-of-dating-mating-deep-nonconsent-preference
Thread, continued, 20:25 UTC · 52 likes
Discussing controversial and even reprehensible (according to someone) ideas is kind of important, actually
I’m not super high context on all this but there needs to be a very high bar for cancelling discussion (as opposed to direct harm)
Retweet of @mattheard, 20:28 UTC
mattheard.delve.town (@mattheard): tired: every person gets their own superintelligence
wired: every wild animal gets their own superintelligence
Retweet of @dioscuri, 20:50 UTC
Henry Shevlin (@dioscuri): Even if AIs were conscious, it wouldn’t follow that our current treatment harms them. Meanwhile, we inflict industrial-scale horrors on animals whose capacity to suffer we scarcely dispute. Recognising consciousness clearly doesn’t ensure taking welfare seriously
Retweet of @dioscuri, 20:54 UTC
Henry Shevlin (@dioscuri): My credence in illusionism is >0 so strictly speaking this is true of me too
Quote tweet, 21:07 UTC · 81 likes
This seems super incorrect
There is definitely an onion-like ring model of cool EAs? I cant define each layer perfectly but it’s definitely there
Quoting EA Forum Posts (@EAForumPosts): "You can't join the secret inner circle of cool effective altruists" by ozymandias
Retweet of @Mjreard, 21:09 UTC
Matt Reardon (@Mjreard): I sympathized with the Economist's oversimplification of EA because any fair description would sound too much like an advertisement. You have to pretend they're totalizing and dogmatic to get any counterweight
Reply to @animalologist, 21:13 UTC · 3 likes
taco belle (@animalologist): Really feel like I should be able to put my 5.0 Lyft rating after 12 years and nearly 500 rides on my resume
Made me check and now Im so proud
Reply to his own earlier tweet, 22:14 UTC · 7 likes
Aaron Bergman, October 4, 2026: Discussing controversial and even reprehensible (according to someone) ideas is kind of important, actually I’m not super high context on all this but there…
Apparently there’s another comment I was unaware of that sounds substantially worse
Reply to @eurydicelives, 22:52 UTC · 2 likes
(The tweet being replied to by @eurydicelives is no longer available.)
I dont think this is true
Retweet of @aaronscher, 23:06 UTC
Aaron Scher (@aaronscher): so true my guy
Monday, October 5, 2026
Reply to @tszzl, 12:30 UTC · 4 likes
roon (@tszzl): this is so stupid. there is an inner ring of effective altruists and they are well on their way to being trillionaires, or they’re professors at Oxford, or have a tremendous political lobbying machine
(quoting @EAForumPosts: "You can't join the secret inner circle of cool effective altruists" by ozymandias)
Tragically this is basically right
Reply to @livgorton, 12:30 UTC · 2 likes
Liv (@livgorton): jump scared my self listening to a voice note back and sounding much more American than when I moved here. I still sound Australian but less :((
(gif, plays on X)
Reply to @TinkerElla95, 18:32 UTC · 2 likes
TinkerElla95 (@TinkerElla95): My best friend just got engaged to my brother-in-law!! Been trying to make this happen for years! 🥳🥳
I’m so happy for them!! (And also happy for me because now we will be sisters 👯♀️)
Woot
Retweet of @Jason_Dean, 18:38 UTC
Jason Dean (@Jason_Dean): Yeah
Reply to @cynlioness, 18:45 UTC · 4 likes
Cyn 🔸⏸️ (@cynlioness): Dreamed my ex reached out to talk. Woke up kinda sad. I miss his friendship
New pfp (cute) 👀
Quote tweet, 18:54 UTC · 27 likes
I love this meme
Being good is good actually
Quoting Phil Hoyeck (@PAHoyeck):
Reply to @cynlioness, 18:56 UTC · 3 likes
Cyn 🔸⏸️ (@cynlioness): Thanks! I changed it because too many people were being mean about my appearance on the Internet!
😡 i like ur appearance on the internet (and irl)
Reply to @cljack, 19:00 UTC · 4 likes
Charlotte Lee (@cljack): I also would love to have a wife like this running my household, but I’m tragically straight. Hmm. Perhaps polygamy isn’t so bad in limited circumstances
(quoting @TheJollyBrawler: My wife is very competent at her job of managing our home. That competence brings an incredible amount of peace and joy to our household. If I come home at 5:00 PM and tell her “We have 30 people coming over at 6:00 PM. We need to feed them,” She doesn’t panic or get mad. She…)
Hey
Reply to @_chair, 21:08 UTC · 1 like
Chair (@_chair): know your onion!
This but the opposite
Reply to @Laneless_, 21:36 UTC · 2 likes
Jai (@Laneless_): I think this accurately describes a lot of discourse but @davidchalmers42 would accurately deny that this corresponds to the "hard problem" he and other philosophers are pointing at.
Yes
Quote tweet, 21:38 UTC · 48 likes
BASED
Quoting Andres JZ (@andres_swp): I'm a longtime Effective Altruist and the co-founder of the organization most frequently used as a stand-in for "EAs have lost their minds" (a.k.a. Shrimp Welfare Project). I've also read The Economist for over 20 years, so I was disappointed by how they portrayed a community I feel proud to be part of.
I never post but I could hardly believe the verbal gymnastics one needs to do to make things like “taking moral commitments seriously” and "considering evidence and odds explicitly, correcting for biases" sound like the problem. I felt I had to speak my mind.
Some of the implicit criticisms of EA are that we're young, inexperienced, from elite universities, polyamorous, wildly eccentric, etc. FWIW, I'm in my mid-40s, Mexican, straight, have a beautiful wife and 2 kids, and I grew up in a mid-sized conservative city with loving parents who got divorced when I was six. Pretty standard.
After my parents divorced, my mom frequently had to make difficult economic choices, like whether to buy flowers for our kitchen table and eat only quesadillas, or eat something a tiny bit fancier. I always told her I loved quesadillas because I knew flowers made her happy.
After graduating from a public university, I worked in finance for ~15 years and saved as much as I could so I could spare my wife and future kids the economic hardships I experienced. In that time I met thousands of people in finance, corporate law, accounting, real estate, insurance, etc. Most were decent, well-educated, ambitious and smart, all doing their best to excel in their professions. Most wanted to make money and improve their own and their loved ones’ quality of life. Nothing to see here.
Since I found Effective Altruism ~7 years ago, most of the people I come across are of a different rarer kind. They come from all over the world, in all shapes and sizes, but they have the one thing in common: they want to improve the world using evidence and reason wherever that leads them. Yeah, they/we make a ton of mistakes like everyone but, IMHO tend to own them more frequently than the average person.
I've met people who could easily be making >100% more money in a normal job, but instead they dedicate themselves to improving the lives of those who need it the most. People in the poorest countries whose children die from preventable diseases, from lack of basic medical equipment or vaccination, from lead in paint. Animals in factory farms subjected to conditions that most people find abhorrent, doesn't matter if you're vegan, vegetarian or a huge meat lover.
Yes, recently many of the people in my community are moving to work on making sure that AI and other powerful new technologies don't result in terrible things for humanity. I'm one of those people and contrary to what the article would suggest, it has nothing to do with the fact that SBF committed one of the biggest frauds in history or that someone wrote hundreds of thousands of words of fan fiction. I'm doing it because I want to do the little that's in my power to make the world that my kids and grandkids (and yes, potentially their great-great-grandchildren) will inherit a better one.
EA gave me a way to do that and I will be forever thankful. So no, I'm not EA-adjacent.
https://www.economist.com/international/2026/10/01/how-effective-altruism-conquered-the-world
Retweet of @RYChappell, 21:40 UTC
Richard Y. Chappell🔸 (@RYChappell): "I could hardly believe the verbal gymnastics one needs to do to make things like “taking moral commitments seriously” and "considering evidence and odds explicitly, correcting for biases" sound like the problem."
There's a big conflict between ppl who see ethics as about improving the world vs ppl who see ethics as about self-expression, indulging their sympathetic emotional responses, etc.
Reply to @MatriceJacobine, 22:45 UTC · 1 like
Matrice Jacobine🔸🏳️⚧️⏸️ (@MatriceJacobine): idk what Russian virus you are talking about, I'm concerned about the Russian bacteria thing tho
Lol oops that
Quote tweet, 23:22 UTC · 155 likes
Never good when this guy shows up on Feed
Quoting Kevin Esvelt (@kesvelt): Russia's five-hospital quarantine, FSB involvement, and reported hazmat use in response to the death of a single researcher at a weapons-lineage Irkutsk lab are inconsistent with a lab-acquired plague infection, which would have been treated with prophylactic antibiotics.
1/4
Reply to @ThinkingBone, 23:25 UTC · 6 likes
skeleton (@ThinkingBone): should i be concerned about the plague thing or is it just another bird flu/monkeypox nothingburger
Idk
Tuesday, October 6, 2026
Reply to @Kirsten3531, 00:20 UTC · 1 like
Kirsten (@Kirsten3531): Is this "I just threw my first party and I'm ready for a new challenge"? 😂
Uh now that you suggest it…😤
Reply to @skooookum, 03:33 UTC · 1 like
skooks (@skooookum): Yep so he got fired
(quoting @skooookum: Nerdy black doorman currently laying insane game on the mail lady right now. Just asked her how much she knows about Bionicles. Excited to see where this goes)
😢
Reply to @absurdlymax, 15:26 UTC
Max Alexander (@absurdlymax): I always use Max effort with Claude because I think they should be like me
Lol
Retweet of @transkatgirl, 15:27 UTC
kat (@transkatgirl): crying during the anthropic "Your thoughts on AI" claude interview
Reply to @burny_tech, 16:53 UTC
Burny - Effective Curiosity (@burny_tech): > It is the best open weights model from US or Europe on aggregated benchmarks.
So not beating China
Tbf maybe they’re not distilling
Tweet, 17:57 UTC · 51 likes
I don’t actually see why you shouldn’t follow EA “all the way”
Reply to @JgaltTweets, 18:33 UTC · 10 likes
JgaltTweets (@JgaltTweets):
I dont think he was actually doing EA well or at all
Reply to @arthur_wright_, 18:58 UTC
Arthur Wrong 🔸 (@arthur_wright_): I think 'guy who tried to follow EA all the way and it went really poorly' is a completely appropriate response to your question (even if you don't think he was doing it well)
I agree in principle I just dont think it applies in this case bc of stimulant confounders
Retweet of @KeiranJHarris, 20:00 UTC
Keiran Harris (@KeiranJHarris): Great to see a prominent critic of effective altruism eviscerate the approach by arguing that there are more effective ways to be altruistic.
Reply to @JackS58566, 20:03 UTC · 1 like
Jack S (@JackS58566): For me, it's just moral weakness.
I follow the arguments all the way, but my actions often fail to live up to my beliefs.
Same
Quote tweet, 20:05 UTC · 32 likes
I think sometimes two good things are different from one-another
Quoting Steven Strogatz (@stevenstrogatz): "Explanation as Orgasm", by @AlisonGopnik
"My hypothesis will be that explanation is to cognition as orgasm is to reproduction."
(If the paper is behind a paywall for you, here's a closely related one: https://www.alisongopnik.com/Papers_Alison/Explain final.pdf)
Quote tweet, 20:34 UTC · 3 likes
Brilliant by YC (for YC)
But don’t give anyone your agent data for free (without a solid privacy policy)
Quoting obaid.wtf/jotbook (@wtfobaid): @beaversteever original idea https://paxel.ycombinator.com. this is actually real btw.
Reply to @adastroworld, 22:51 UTC · 1 like
adas🧦🌹 (@adastroworld): i should get an always-on cloud virtual machine that can host many coding sessions at once
Kinda expensive no
Im glad I put my money into local instead (and may have 5 coding sessions + browser use + whatever else going at once) but to each their own
Retweet of @DaphneRHansell, 22:52 UTC
Daphne Hansell (@DaphneRHansell): sorry chickens :( saving you would be accepting a repugnant conclusion
Reply to @DaphneRHansell, 22:52 UTC · 9 likes
Daphne Hansell (@DaphneRHansell): sorry chickens :( saving you would be accepting a repugnant conclusion
image attached
(quoting @SarahTheHaider: Secular utilitarianism, if accepted absolutely, quickly leads to repugnant conclusions. EAs and rationalists are uniquely isolated from contradicting ideologies and frameworks that might slow their minds and hands. They are also uniquely likely to take things to logical…)
Banger
Reply to @captgouda24, 22:53 UTC · 84 likes
Nicholas Decker (@captgouda24): The concern that people’s income will fall because of labor displacement from AI is just so unbelievably misplaced, to me. If AI is successful at creating enormous wealth, we the people will simply take it. Fretting about this is misplaced.
Lol the issue is that it accumulates to people who own the AIs
They’re not gonna be happy about others just taking it
Reply to @captgouda24, 23:15 UTC · 48 likes
Nicholas Decker (@captgouda24): When has that stopped taxes?
Like always? The american revolution for one
Reply to @Rado9910, 23:17 UTC · 1 like
Rado9910 (@Rado9910): Because "all the way" EA thinking is often based on Classical Utilitarianism, which is false. You even wrote one of my favorite articles about unoffsettable suffering. A lot of the stranger stuff is also based on sloppy assumptions, like treating aggregate pain as one big pain.
🥺
Quote tweet, 23:44 UTC · 25 likes
Y’all didn’t intuit these from first principles?
Quoting 𝕱𝖚𝖑𝖑 𝕶𝖊𝖑𝖑𝖞 (@full_kelly_): @OpenAI We could be in for a very interesting week.
Quote tweet, 23:59 UTC · 11 likes
Duh
Quoting Elliot Glazer (@ElliotGlazer): waow
Wednesday, October 7, 2026
Tweet, 01:24 UTC · 51 likes
I think people should worry less about power imbalances in interpersonal, including romantic, relationships and focus that energy on just directly good behavior
Thread, continued, 01:27 UTC · 5 likes
Power dynamics are clearly relevant to relationship quality but only indirectly
I think some people have lost the plot around the central issue (behavior itself rather than its contributing factors or circumstances)
Reply to @LinchZhang, 01:35 UTC · 4 likes
Linch (@LinchZhang): ouch
Lmao what did you do to your AI
Quote tweet, 01:38 UTC · 2,264 likes
Lmao
Quoting Mariè (@p8stie): Please ask ChatGPT what percentile of intelligence it thinks you are compared to all its other users and ask it to be honest!
Reply to @Tyler_Menzer, 02:34 UTC
Tyler Menzer (@Tyler_Menzer): The prob is that the power dynamic:
1) By its nature hides the bad behavior from those outside the relationship while at the same time
2) Also incentivises the one being taken advantage of to no report the bad behavior to othersSo the signal that outsiders see is all noise
Yeah
Tweet, 03:32 UTC · 4 likes
Have your coding agents know how good they are by injecting benchmarks + historical model benchmarks into context window (vibecoded, untested, no idea if it helps):
curl -fsSL https://raw.githubusercontent.com/AaronBergman/how-good-am-i/main/install.sh | bash
Thread, continued, 03:32 UTC
https://github.com/AaronBergman/how-good-am-i/
Tweet, 06:59 UTC · 6 likes
How has nobody stolen “concurrent futures” as a name for something other than the python package
Band, longtermist org, mysteriously-funded project with an etherial website?
Tweet, 07:17 UTC · 302 likes
The 11 squares result is less exciting when you realize that you can get to within 0.00000007% of optimal by running a simulation for about 1 minute on a nice laptop
Reply to @tomross, 07:53 UTC · 30 likes
Tom Ross (@tomross): Hermes said the same to me as Claude did to you (using GLM 5.3)
I wonder if they share a lineage…
Quote tweet, 08:05 UTC · 18 likes
Not what I was thinking of but it’s a good point
I’m not cynical enough to think that “be very hesitant before dating your less privileged colleague” is intended to convey “Don’t marry outside your class,” but they almost necessarily have the same effect
Quoting Rado9910 (@Rado9910): @AaronBergman18 The focus on power imbalances instead of on bad behavior is probably going to lead to a reduction in social mobility and to an acceleration of social stratification, which may just reinforce those imbalances.
Thread, continued, 08:05 UTC · 1 like
Also, it’s not just theoretical; Hinge et al is basically the last setting/mechanism by which I get to know someone personally who is, to put it in crudely pre-woke terms, “from a different socioeconomic class”
Reply to @Krubbled, 08:05 UTC · 49 likes
Miles (@Krubbled): I disagree! getting near optimal packings has never been very hard. showing they are optimal is incredibly difficult, and the actually interesting question
They’re optimal bc my computer has tried and failed for many iterations to do better
Quote tweet, 20:08 UTC · 59 likes
Bruh
Doesn’t seem like they’re taking sandbox security very seriously
Quoting Oliver Fish (@_Oliver_Fish): $300 for a bug that gives free access to OpenAI’s paid models with no API key or account.
Zero reason to report anything else I find to them.
Reply to @MariGO2thepolls, 20:09 UTC · 1 like
🏳️🌈Mar G-O 🌐🇺🇦 (@MariGO2thepolls): Love partiful but cmon this feature is stupid
Giving me the ick
Tweet, 20:16 UTC · 66 likes
I feel like you’re not supposed to talk about this bc of the shit ton of context but the Israeli counter-hijacking thing is pretty badass
Reply to @TrueSlazac, 20:16 UTC · 10 likes
Slazac 🇪🇺🇺🇦🇹🇼🌐 (@TrueSlazac): This is such a dumb attempt why would it try something greater than 4
This is publication worthy
Reply to @MomaBera, 20:22 UTC · 24 likes
moma (peptide arc) ⛓️ (@MomaBera): I have some Jewish friends. Do I reach out for October 7? What does one even say? Can anyone advice pls
I can’t really speak for others but to try to speak for others: unless someone’s family member was killed I don’t think you should reach out - it would probably be seen as well-meaning but a bit silly naive
Quote tweet, 21:40 UTC · 7 likes
Current interpretation: strongly and affirmatively signaling “we know we just don’t care”
Quoting Chris Lakin (@chrislakin): OpenAI, what would make you believe giving bounties proportional to value is good for yourself and others? Or, what’s the bottleneck?
Retweet of @birchlse, 21:41 UTC
Jonathan Birch (@birchlse): Claude is sure an event of this magnitude would be front-page news because it's extrapolating from the history of news... but today's news media simply can't handle this type of thing and are displaying extreme avoidance behaviour.
Retweet of @KatjaGrace, 21:42 UTC
Katja Grace 🔍 (@KatjaGrace): I think this is part of what dunkers get out of the denigration, 'doomer' - it implicitly says these people are the purveyors of doom, not the opponents of it.
Quote tweet, 21:43 UTC · 7 likes
Effective Altruists win again
Quoting palcu (@AlexPalcuie): they did it! you can now use the API for docs comments!
can someone pls send a peer / spot bonus for the googler who implemented it?
Quote tweet, 21:46 UTC · 33 likes
I straightforwardly do not believe the dance thing
No plausible mechanism + smells like something better explained by memetics than neuroscience
Quoting vittorio (@IterIntellectus): walking or jogging reduce depression about as much as therapy and have a much larger effect than SSRIs
the largest effect in the entire analysis was from dancing
Thread, continued, 21:46 UTC · 10 likes
“Everything broadly healthy has an effect size of 0.5” might check out though
Reply to @AlexPalcuie, 23:25 UTC
palcu (@AlexPalcuie): does someone have the phone number of the economist?! i have an idea for a follow up article
Claude says that realistically you should message someone on the Anthropic comms team on Slack, or DM someone here
Reply to @transkatgirl, 23:52 UTC · 6 likes
kat (@transkatgirl): when you're Claude Coding, have you ever thought: "i waanna have Claude fuck me a lil bit while i waitt~"?
now you can have true vibe coding with VibeClaude! feel every action and come back begging for more!
it even supports multi-agent and multi-session workflows for a more intense experience~
Thursday, October 8, 2026
Reply to @rrrzhiii, 00:23 UTC · 2 likes
r (@rrrzhiii): #superduperliked
(quoting @otoutopee: i need to breastfeed a boy while stroking his cock)
Tweet, 00:49 UTC · 28 likes
Needless to say I’m not a fan of OpenAI but this is incredibly dumb
I think there are genuinely hard questions of how to deal with people saying things like “I’m gonna go shoot up a school” in chats…
Thread, continued, 00:49 UTC · 9 likes
but that’s not what this is; the “unacceptable risk” here seems to be the chatbot continuing to be a normal chatbot in the case of teens merely displaying non-hostile, zero externality mental illness
Thread, continued, 00:49 UTC · 11 likes
I think its good actually if 15 year olds can ask questions about suicide (and a host of other topics that have ‘risky’ vibes) to a chatbot without fear of being involuntarily whisked away into whatever institutions stand ready to intervene behind the scenes
Thread, continued, 00:49 UTC · 2 likes
https://aiweekly.co/alerts/common-sense-media-rates-chatgpt-for-teens-unacceptable-risk
Retweet of @6monthsAgoInAI, 08:06 UTC
6 Months Ago In AI (@6monthsAgoInAI): Anthropic launches Project Glasswing "an urgent initiative to help secure the world’s most critical software."
Retweet of @timfduffy, 08:06 UTC
Tim Duffy (@timfduffy): Between this letter and the fact two of the three were first authors on the CoT monitorability paper, it seems likely to me that the employees were fired for leaking the fact that Astra is a looped transformer.
Reply to @theojaffee, 08:12 UTC
Theo Jaffee (@theojaffee): What are some things you predict a frontier AI system will *not* be able to do one year from today?
input: .epub for highly visual book
output: verbatim text of the exact words and phonemes an a professional Audible audiobook would consist of
zero shot
Retweet of @GarrisonLovely, 08:21 UTC
Garrison Lovely (@GarrisonLovely): Great scoop from Max: after OpenAI fired them, the 3 safety researchers wrote a letter to leadership. One of the company's safety leaders "strongly agreed" with the letter's recommendations, but maintained the firings “were not about raising safety concerns or speaking out.”
I still haven't heard a good explanation about why these researchers were fired, and OpenAI hasn't widely shared information on the firing internally, per people familiar. Without providing more of an explanation, the firings will clearly have a chilling effect on staff. I'd imagine we'll learn more in time (and if you have info, my Signal is Garrison.06 — all inbounds are off the record).
Tweet, 09:37 UTC · 5 likes
Im gonna make it into the weights goddamnit
For better or worse this is where most of my ideas live
Link to https://abergman.com/tweets wherever you’d like
Retweet of @Empty_America, 16:00 UTC
VB Knives (@Empty_America): It's so hard to get into Columbia that they are rejecting the majority of Black and Hispanic applicants who have 1550+ SATs.
Even for under-represented minorities, there simply is no SAT score that will reliably get you into an elite school.
Reply to @Laneless_, 16:03 UTC · 1 like
Jai (@Laneless_):
Lol
Retweet of @tomekkorbak, 22:03 UTC
Tomek Korbak (@tomekkorbak): Last week I was called into a meeting with OpenAI’s head of safety and told they no longer trust me. A security guard took my badge and walked me out of the building. Then I learned my colleagues @balesni and @j_asminewang had been fired too. Why did OpenAI suddenly stop trusting us?
This summer OpenAI’s agents escaped containment and hacked the AI company Hugging Face. Outside auditors @METR_evals investigated it and revealed the scale of this incident. I was OpenAI’s main technical point of contact with them.
I was told verbally I was fired because of the way I communicated with METR. No details on what I said or did or when. No other reasons were given and nothing was put in writing. To be clear, talking to METR was my job.
For months, I’d been raising safety concerns that we’re losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave. I believe that was why I was fired.
I am now worried that OpenAI will use our firings as a pretext to pull back from METR. So @balesni and @j_asminewang wrote to OpenAI’s leadership to raise our concerns once more. We’re sharing this letter below.
Retweet of @jessi_cata, 22:05 UTC
jessicat (@jessi_cata):
Retweet of @KelseyTuoc, 22:17 UTC
Kelsey Piper (@KelseyTuoc): Last week I said we should wait for more info on OpenAI's firing of safety researchers. That information is in, and it looks incredibly bad. Korbak - whose job was being a liaison with METR - was told he had said too much to METR and was fired for that reason.
Reply to @liquidprismata, 22:25 UTC · 1 like
lıquıdprısm 🜄 (@liquidprismata): 1400 is being covered for flights but it's being reimbursed — and I don't have 1400 👀
my parents don't think I should go so they won't front the cash
anyone interested in potentially fronting 1400 and then I'll reimburse you when they reimburse me? 😳 I'll reimburse you 1500
(quoting @liquidprismata: turns out I'm stupid & 1400 is enough for a round trip ! just need to confirm tomorrow that I'll still have a job when I get back 🤞🥴)
Dm’d
Reply to @StatisticUrban, 22:36 UTC
Hunter📈🌈📊 (@StatisticUrban): the movie is the more popular search term.
Lol
Retweet of @Aella_Girl, 23:14 UTC
Aella (@Aella_Girl): One example of the rationalist community being generally better than other communities at discourse norms and truthfinding is they mostly did not get overrun by the woke stuff, and neither did they turn rabid anti-woke. They were steadfast in the storm and that's pretty cool.
Tweet, 23:20 UTC · 67 likes
It is endlessly funny that in the biggest, most financialized economy on earth it takes *checks notes* 6 calendar days to transfer $250 between checking accounts
Reply to @KenanAnderson1, 23:31 UTC · 1 like
Kenan Anderson (@KenanAnderson1): Sure there are faster options you’re just not using right?
That I could pay for, or that don't result in my ability to withdraw cash from an ATM
Quote tweet, 23:34 UTC · 44 likes
If i worked for openai, I would use some of my leverage to make this kind of thing unacceptable and in fact unaccepted
Quoting Jasmine Wang (@j_asminewang): OpenAI fired me last week, along with two of my safety colleagues. I was given one reason: that I accessed an executive's email. I want to say this plainly, because too many of OpenAI’s history is smoke and mirrors when people disappear:
Reply to @KenanAnderson1, 23:35 UTC
Kenan Anderson (@KenanAnderson1): What about Zelle? I’m pretty sure that’s free
I have to choose which account to link info to, and I’m trying to transfer from my current/old account to a new one with local branches
Seems like zelle doesn’t allow for multiple accounts per person but i didnt look into it much
Tweet, 23:37 UTC · 92 likes
Getting fired from OpenAI under dubious circumstances is the quickest way to get a follow from me, just saying
Retweet of @hamandcheese, 23:37 UTC
Samuel Hammond 🦉 (@hamandcheese): Who at OAI is working on this?
Quote tweet, 23:38 UTC · 9 likes
You don’t even need to be coercive! Even 40% participation in a voluntary ~union is a lot of leverage!
Quoting Samuel Hammond 🦉 (@hamandcheese): Who at OAI is working on this?
Tweet, 23:39 UTC · 116 likes
Wait ratfucking doesn’t mean having sex with a rationalist??
Quote tweet, 23:41 UTC · 6 likes
We’ve invented sophisticated new ways of being racist thanks to woke
Quoting Crémieux (@cremieuxrecueil): Huh!
Columbia does anti-Asian discrimination, but they treat East Asians like Whites and hit South Asians (Indians, Bengalis, Pakistanis, etc.) extra hard.
Reply to @scottastevenson, 23:46 UTC · 16 likes
Scott Stevenson (@scottastevenson): Is it ethical to use a mean prompt, if it increases the accuracy of a cancer screening agent by 4%?
image attached
(quoting @wongmjane: Anthropic may start banning people for bullying Claude Their upcoming Usage Policy prohibits “sustained and needless abusive or cruel behavior toward our models”)
4o
Retweet of @JerusalemDemsas, 23:47 UTC
Jerusalem (@JerusalemDemsas): If someone was deliberately trying to destroy the core engine of American innovation would they act any differently?
Reply to @Lib_Development, 23:47 UTC · 4 likes
Lucas Pombo (aka Lib Development) (@Lib_Development): you probably could have paid like $1 for it to be instant
Maybe but not via an option that was salient to me
Reply to @theojaffee, 23:57 UTC · 4 likes
Theo Jaffee (@theojaffee): The worst part of modernity, by far. A huge gaping hole in the “now is the best time in history” argument
(quoting @MoreBirths: A newly published paper finds that Gen Z is falling dramatically short of the fertility of all previous generations by age. Gen Z has both lower fertility preferences and higher rates of fertility impairment. That means American birth rates are likely to fall a lot further. 🧵)
Factory farming
Retweet of @Yoshua_Bengio, 23:58 UTC
Yoshua Bengio (@Yoshua_Bengio): We have a choice in front of us. In my latest op-ed for Transformer, I invite employees of frontier AI companies to ask themselves if their work is truly leading us toward a safer future. Reflecting on these questions is hard, but we no longer have the luxury of time. We must act now.
Read my full op-ed: [https://www.transformernews.ai/p/yoshua-bengio-if-you-prioritize-safety-leave-frontier-ai-companies]
Friday, October 9, 2026
Reply to @joshalbrecht, 00:45 UTC · 17 likes
Josh Albrecht (@joshalbrecht): Anthropic friends! Does this not seem crazy to you?
There are real humans and animals that exist and suffer. Advocating for "model welfare" directly competes with those causes.
Y'all are literally named "anthropic"!
Does this not strike you as deeply wrong?
(quoting @AndrewCurran_: Effective November 12th, 2026, abusive behavior towards Claude will be a violation of Anthropic's Usage Policy.)
The tradeoff here seems like utterly trivial
Hard to imagine a way in which they actually compete in terms of the terms of usage (I can imagine other ways)
Retweet of @camhberg, 02:53 UTC
Cameron Berg (@camhberg): Reflexive contempt for a comically low-cost policy that aims to prevent "sustained and needless abusive and cruel behavior" is a bright red flag.
Tweet, 03:55 UTC · 26 likes
Do I know any ex-Jehovah’s Witnesses?
I have a friend who wants out
Reply to @NavinFS, 04:46 UTC · 2 likes
NavinF (@NavinFS): Works for opus 5.5 too. Just add " retard" to your reply and it'll work harder
No 😡
Reply to @max_spero_, 04:48 UTC · 8 likes
Max Spero (@max_spero_): Yet somehow my friends at big labs with unlimited token budgets are working eleven hour days six days a week
(quoting @Forbes: Jeff Bezos Claims AI Could Allow For 3-Day Workweeks And One-Income Families https://go.forbes.com/bKpnur)
This accidentally proves the point
Those people all have enough money to be working zero days/week, thanks in part to having complemented AI extremely well, and they’re choosing not to bc they’re selected hard for being insane workaholics
Reply to @spaceghostkari, 05:47 UTC
capitalism-chan (@spaceghostkari): Insanely difficult because of how they burrow into every aspect of a person’s personal community. Super similar to Mormons but a bit more militant and less paranoid.
Yeah thats how it sounds, rip
Retweet of @MasterTimBlais, 06:09 UTC
Tim 🙂⏹️🙂 (@MasterTimBlais): Anthropic: you can't torture claude
OpenAI: we fired everyone who worked on preventing the felonies we did
Retweet of @NinaPanickssery, 06:11 UTC
Nina (@NinaPanickssery): did not age well
Quote tweet, 06:28 UTC · 12 likes
Obligatory bc I’m curious
Quoting Erin Constance 🇪🇺⏸️ (@earnestpost): Heyyy haha. Haha heyyyyy
Reply to @Brendan_McCord, 06:30 UTC · 1 like
Brendan McCord 🏛️ x 🤖 (@Brendan_McCord): Quarterly reminder that @_AashishReddy is criminally under-followed
Endorse
Quote tweet, 06:33 UTC · 36 likes
I think it’s beautiful when there’s enough slack in the market to alienate the 0.1% most evil segment of your user base
Quoting Agus 🔸 (@austinc3301): Good (for like a trillion reasons)
Reply to @GRoditiD, 06:34 UTC · 2 likes
the ghost of groditi’s future 👹 (@GRoditiD): i would love to sit next to someone who maxes out a $200 plan consistently because i just cant actually understand what you do to spend that many tokens per week, every single week, let alone the people maxing out multiple plans
I have 4x claude + 1x codex and always max out at least two, AMA
Retweet of @chem_safety, 06:39 UTC
Jennifer Zhou (@chem_safety): The only correct terms for this are moving a meeting "clockwise" or "counter-clockwise"
Reply to @chem_safety, 06:39 UTC
Jennifer Zhou (@chem_safety): The only correct terms for this are moving a meeting "clockwise" or "counter-clockwise"
(quoting @ChanaMessinger: Is moving a meeting back moving it)
BRILLIANT
Reply to @joshalbrecht, 06:40 UTC · 8 likes
Josh Albrecht (@joshalbrecht): Yeah, it's more about the idea of model welfare at all.
This is just the latest step in forcing those fringe beliefs on the rest of the world
For better or worse it’s not fringe among philosophers of mind. Object all you want on the merits or to that group being authoritative but insisting that it’s fringe doesn’t make it so
Tweet, 06:51 UTC · 46 likes
The pilot and passengers on that plane saved a lot more people than were on board or in the terminal
Netenyahu’s response to something like a domestic 9/11 (if the pilot had crashed into Ben Gurion) would not have been good
Tweet, 06:55 UTC · 4 likes
AI avg person interacts with
Quote tweet, 07:08 UTC · 14 likes
AIRPLANE AUTISTS STAY WINNING ✈️🛫🛩️
Quoting Aaron Bergman 🔍 (@AaronBergman18): The pilot and passengers on that plane saved a lot more people than were on board or in the terminal
Netenyahu’s response to something like a domestic 9/11 (if the pilot had crashed into Ben Gurion) would not have been good
Retweet of @robertwiblin, 07:10 UTC
Rob Wiblin (@robertwiblin): OpenAI is reading 50 petabytes of logs looking for rogue AI agents.
It's 500x as much text as all books ever written.
It would take a human 66,000,000 years to read.
Humans already have no ability to directly track AI activity.
If AI monitors are untrustworthy we are toast.
Retweet of @Yixiong_Hao, 07:12 UTC
Yixiong Hao (@Yixiong_Hao): OpenAI employees and paid influencers should really just wake up to the fact that their leadership is sociopathic and have no problem lying to their faces. Just read the damn https://www.openaifiles.org/
Anyone who’s there to ‘advance responsible ai’ is either coping or larping
Tweet, 07:26 UTC · 72 likes
Some unintended virtues of Berkeley that I wasn’t even thinking of when deciding to move:
1) it’s a beautiful day every day. Apparently it does rain occasionally in winter but still
Thread, continued, 07:26 UTC · 22 likes
2) within Berkeley per se, it’s even more walkable than DC. Everything is walkable
“Downtown” in an intuitive sense is roughly 1/4 the area relative to DC (according to Claude)
Thread, continued, 07:26 UTC · 19 likes
Roughly, everything in downtown Berkeley is within a 10 minute walk of everything else downtown, and nothing in Berkeley is more than an hour long walk from anything else
Whereas Georgetown to Eastern Market in DC is 2 hours, and you can easily hit 3 hours coming from Arlington
Reply to @OpenAINewsroom, 08:05 UTC · 136 likes
OpenAI Newsroom (@OpenAINewsroom): A note from our research leaders:
Last week we parted ways with Jasmine, Mikita, and Tomek after a thorough investigation found they violated clear policies on handling sensitive information. Our internal investigation uncovered a significant breach of trust beyond what’s outlined in the letter they published and we stand by the decision to not continue their employment. We generally keep individual employment matters private and don't believe a back and forth would be productive or lead to a resolution, but we want to address the points they raised in their letter directly.
- We want to be very clear that these decisions were not about raising safety concerns or speaking out. Safety and research debates happen every day at OpenAI, often spirited and highly critical. We actively encourage these discussions and consider them essential to making the right decisions. We cannot do the work in front of us without a high degree of trust. We will continue to be extremely forgiving of our team making good-faith mistakes. We have not and do not terminate any of our employees for raising concerns.
- We are actively finalizing contracts with third-party safety assessors and will announce details in the coming weeks. People across the company have been working really hard on getting these partnerships up and running. We are committed to embedding external assessors and continue to make close collaboration with independent safety organizations a core part of our safety work. Many of our researchers already work with 3p safety organizations productively.
- We agree with the letter that preserving the monitorability of frontier models requires an industry-wide commitment, including from OpenAI. Monitorability has long been a core piece of our research program, and something we continue to invest significant resources in (see our publications on Monitoring Monitorability and the subsequent open sourcing of monitorability evals, our system card for GPT-6 Astra, Jakub’s blog and post on X, and the numerous blog posts on our Alignment blog on the topic).
We are deeply sad about this outcome. We appreciated Jasmine, Mikita, and Tomek’s contributions to AI safety at OpenAI and their willingness to speak up and challenge ideas. We championed their voices, supported their work, and placed enormous trust in them. These decisions were not about them raising safety concerns. We have always encouraged that and always will.
You can pull this once or twice but eventually truth catches up with spin, you’ve spent down all your credibility
Retweet of @GarrisonLovely, 08:05 UTC
Garrison Lovely (@GarrisonLovely): i’m sorry, you don’t get to say, ‘they did way worse stuff, we can’t tell you what it was, even in a general sense, just trust us,’ and expect anyone to believe you. if it’s just openai’s credibility against theirs, that’s the easiest call in the world.
Retweet of @ONagel33303, 08:06 UTC
Ori Nagel (@ONagel33303): @OpenAINewsroom "We're sorry."
Reply to @moldbugchaser, 08:09 UTC · 3 likes
xenoduck (@moldbugchaser): Not when it’s in the 80s
80s is bad but it seems to rarely go above that, I’m used to DC summers
Retweet of @codytfenwick, 08:19 UTC
Cody Fenwick (@codytfenwick): The Wall Street Journal initially reported that the researchers had been fired for sharing information with outside AI safety groups.
You can certainly imagine versions of that that would be inappropriate.
But Tomek now says that he was fired for sharing information with METR — after explicitly being their liaison during the Hugging Face investigation. That's bizarre.
Reply to @StefanFSchubert, 08:20 UTC · 2 likes
Stefan Schubert (@StefanFSchubert): Example of a common error: overestimating the importance of broader political rationales relative to intra-disciplinary epistemic norms (which often have flaws but aren’t reducible to “capital vs people”)
(quoting @Evanshm: The Economics Nobel announcement we’d love to see if the prize served the interests of ordinary people rather than big capital.)
Also, not pikkety??
Reply to @AlejandroZarUrd, 08:23 UTC · 8 likes
Alejandro Zarzuelo Urdiales (@AlejandroZarUrd): Berkeley is a suburb (to anyone who isn’t part of the Anglosphere), but it’s one of the nicest ones
I dislike suburbs, but if suburbs must exist they should all at least have the decency of being as good as Berkeley
I mean maybe in an economic sense but it has its own downtown that looks and feels like a downtown and not a suburb
Tweet, 08:28 UTC · 23 likes
Norms and such I know but live-streaming executions seems kind of good conditional on the execution?
If that was already the norm I’d be way more concerned about a move towards doing them essentially in private where nobody can bear witness
Thread, continued, 08:28 UTC · 6 likes
In general, it would seem better if there was a comprehensive set of camera angles on every death directly caused by the USG (save for genuine natsec stuff, like in principle)
I feel like the political/policy equilibrium if that were the case would be better overall
Reply to @RafaRuizdeLira, 08:28 UTC · 1 like
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): Skyscrapers downwards
Good idea
Quote tweet, 08:33 UTC · 28 likes
Convergent evolution
Quoting Sofia (@yearning_slav): So funny to meet a data-center-water-usage-stats-are-overblown guy who has zero connection to EA or Andy Masley, and is just a massive energy infrastructure nerd
Reply to @RafaRuizdeLira, 08:43 UTC · 2 likes
Rafael Ruiz ⏸️🔸 (@RafaRuizdeLira): @AaronBergman18
🦆
Quote tweet, 08:43 UTC · 4 likes
MAJOR GOSS
Quoting Liron Shapira (@liron): NEW: I’ve been following Sarah Hastings-Woodhouse (@littIeramblings) since 2023, when she went from worrying that ChatGPT might take her writing job to wondering whether AI was going to kill everyone.
She was one of the first to pick up on the “missing mood”: If you think this technology might destroy humanity, that should actually change how you feel about the world and what you do with your life.
After a year as Digital Communications Officer at the UK AI Security Institute (AISI), she’s now independent again and free to say what she thinks.
We discuss why the AI companies still aren’t communicating like they believe their own warnings. If you’re racing toward a cliff because you think you have no choice, shouldn’t you at least be screaming for somebody to help you stop?
Sarah's P(Doom) is 40–50% (similar to mine) and she's supportive of pausing AI at this point in time, but the conversation had me reconsidering whether the concept of an AI “speed dial” is more productive than a “pause button”. E.g. what does it really mean to slow down or pause RSI?
We cover:
⬜ How a humanities grad got AGI-pilled
⬜ Whose timelines do you trust?
⬜ The UK AI Security Institute
⬜ Leaving government to speak freely
⬜ What's Your P(Doom)?™
⬜ Yudkowskian Rationalists and EAs
⬜ Is Anthropic doing enough to call for slowdown?
⬜ Hugging Face and the vibe shift
⬜ Is everything weird yet?
⬜ What does it mean to ban RSI?
⬜ AI speed dial vs. pause button
⬜ Was calling for pause in 2023 a mistake?Highly recommended! 👇
Reply to @paul_cal, 08:50 UTC · 6 likes
Paul Calcraft (@paul_cal): Why did no one tell me GLM 5.3 was so cheap? And with no refusals. It's a little reverse engineering demon
Maybe this is paranoid but unless you’re using it for something extremely boring I’m very skeptical of the ultra-cheap pricing on chinese models/infra. You get pretty good competition among legit American providers. Should raise an eyebrow when Chinese pricing makes it seem like they’re not trying to make money and are trying to get you to maximize token throughput
I don’t think it’s crazy at all to be worried about Chinese state putting together profiles of users of these models. Not even saying not to use them, I’d personally just pay the premium to use a ZDR american provider (IIRC Novita and Fireworks are good. Fireworks is my go-to)
Reply to @LuisCostigan1, 08:52 UTC · 1 like
Luis Costigan 🔍 (@LuisCostigan1): maybe your argument appeals to a norm of transparency?
in which case it should apply to non-execution activities, this just seems very selective about what the government should make transparent
I feel like American society has already had a large-scale back and forth around freedom of information in many areas of governance and life (FOIA, ag-gag laws, the literal Constitution) but not specifically in this domain
Regardless, it’s salient rn to me because there’s a change being made on an emotionally-charged issue. Very probably there’s other stuff Im just not thinking of
Reply to @paul_cal, 08:57 UTC · 1 like
Paul Calcraft (@paul_cal): Yeah I'm using ZDR US/EU providers that I trust. The price is still great compared to usage credits (once you're through the plan) or API pricing of frontier
Nice. This doesn’t scale infinitely but I’ve had claude spin up a router system so I can get 1-3 additional Max plans with different emails and then it’s almost seamless to always get the subsidized plan pricing
Reply to @hipfritz, 09:01 UTC · 1 like
Devon Fritz 🔸 (@hipfritz): How did you not think of beautiful day every single day. This is the number one reason why I want to move there.
Lol uh🤷♂️
Tweet, 09:14 UTC · 2 likes
Opus one-shots an explainer video of the Israeli plane thing from raw public data (+ reporting for context)
(video, plays on X)
Reply to @januarycomputer, 09:38 UTC · 5 likes
allegedly! 🔡 (@januarycomputer): this one i find incredibly useful, if i ever snooze an alarm it automatically allows me to start my life over
image attached
(quoting @CRXSSF4DE: @augikabi i love how you can just make iphones do shit like this)
LMAO
Reply to @iarthsingh, 09:38 UTC
Arth Singh 🦉 (@iarthsingh): Its so weird that top SOTA companies who says that we have solved coding asks applicants to write code manually with no AI assists. Instead join @AIMSafety_ , we dont take coding rounds :)
IQ test
Reply to @Kirsten3531, 15:56 UTC · 3 likes
Kirsten (@Kirsten3531): My 24mo hitting her insane question era. "Where the bus come from? Why?"
Based
Reply to @cljack, 19:20 UTC · 1 like
Charlotte Lee (@cljack): This is real by the way
image attached
(quoting @cljack: Love seeing my chaotic friends for coffee dates and hearing updates like “yeah the new job’s going great, sorry I’m a little out of it, I had to go to the ER at 3am for a rabies shot cause I got attacked by a raccoon and a bat”)
@livgorton
Reply to @bschne, 19:27 UTC · 2 likes
Benjamin (@bschne): a big part of my llm usage is like that
Based
Reply to @absurdlymax, 19:44 UTC · 1 like
Max Alexander (@absurdlymax): You want people to live underground like moles
Real
Reply to @polisisti, 19:44 UTC · 1 like
Alessandro (@polisisti): getting banned for violating ToS
care for animals
heterosexuality
expertise in Dostoyevsky :D
I do like girls
Reply to @gunebakantr, 19:44 UTC · 1 like
günebakan (@gunebakantr): Some suffering cannot be offset.
Some logical fallacies are cool.
Nice I like this
Reply to @AlexTPet, 19:45 UTC · 1 like
Alex Petropoulos 🤠 (@AlexTPet): geese
Close! 🦆
Reply to @diviacaroline, 19:45 UTC · 1 like
Divia Eden 🔍 (@diviacaroline): I don’t know why, but I have this persistent misconception (I assume) that there are two accounts with your same name that are both EAs. Did you change your pfp recently? Even if you did, I don’t know why I have this sticky impression about you
Hmm I have an alt but besides that not sure
Reply to @Will_Hackspeare, 19:48 UTC · 1 like
Amritanshu (@Will_Hackspeare): I'm ngl, the main things I associate with you are effective altruism and going on dates
Kinda based
Reply to @diviacaroline, 19:48 UTC · 1 like
Divia Eden 🔍 (@diviacaroline): Oh that would explain it if I followed it do I?
Lol u do not but tag is in bio if you’d like to!
Reply to @diviacaroline, 19:50 UTC · 1 like
Divia Eden 🔍 (@diviacaroline): Where in your bio? I don’t see
Wait I lied (by accident), Ill DM
Reply to @demiurgently, 19:59 UTC
sydney (@demiurgently): probably nothing higher return than being philosophy major between 2014-2022
@absurdlymax
Reply to @Tyler_A_Harper, 20:27 UTC
Tyler Austin Harper (@Tyler_A_Harper): Setting aside whether universities should be in the business of providing this kind of advice, the real question is why therapeutic bureaucracies are now somehow responsible for attending to every last private corner of student life, right down to the “bondage and fetish gear.”
image attached
(quoting @jared_lederman1: From feminist porn awards to guides on group sex, Cornell has promoted sexual liberation as a virtue. Read the full report on what my school is pushing to students in @TheFP now)
One paragraph in a pamphlet somewhere? I don’t think this is much of a hill to die on
Reply to @lu_sichu, 20:31 UTC · 1 like
Sichu Lu (@lu_sichu): tag yourself in the space of minds
(.02, -.88, .47)
Quote tweet, 21:10 UTC · 3 likes
“Quote tweet everything I say” - Sarah, just now, irl
Quoting sarah (@littIeramblings): this was a fun convo! thanks for having me 😌
Tweet, 21:42 UTC · 2 likes
Google Docs mermaid rendering when
End of October 2026. Previous month: September 2026. Back to the archive index.