Dustin Moskovitz @moskov.goodventures.org · 29/09/2026seven months later, I am pleased to report that while morale remains alarmingly low, there *has* been a marked reduction in reports of in-group fighting about timelines 3366
Reposted by Dustin MoskovitzMatt Darling @besttrousers.bsky.social · 26/09/2026So what's very funny here is the actual underlying claim is "companies are faking enrollment and just walking away with the subsidies" (which is plausible!( but politicians are unable to actually frame it like this. 7837
Reposted by Dustin MoskovitzEd Maibach @maibached.bsky.social · 25/09/2026Happy Stanislav Petrov Day He passed in 2017. As Encyclopedia Britannica puts it, “He is survived by his two children, two grandchildren, and the entire human race.” If you don't know his story, please take 3 minutes now to learn it. @gchalliance.bsky.social @docsforclimate.bsky.socialsammatey.substack.comHappy Stanislav Petrov DayTake five minutes to not destroy civilization 252097918
Dustin Moskovitz @moskov.goodventures.org · 23/09/2026He pushes back against doomerism while also calling for the labs to be *fully shut down* if lesser harms occur. I really think we have to go slow to keep going fast here - nobody wants the future where a catastrophe leads to a full shutdown but we're painting ourselves into a corner with rhetoric. 3360
Dustin Moskovitz @moskov.goodventures.org · 22/09/2026ASI is officially here, but it's all much dumber than we thought it would be. 5466
Dustin Moskovitz @moskov.goodventures.org · 22/09/2026imagine if all the evidence for the Iran war becoming a neverending disaster was “a bunch of guys played games about war and agreed it was very bad and might happen” instead of empiricism. true strategists know there is only *one* tried and true method for finding out 2589
Dustin Moskovitz @moskov.goodventures.org · 22/09/2026Everyone who is not an EA: EAs suck, esp Yud Yud: Nobody thinks EAs suck more than me! 6912
Dustin Moskovitz @moskov.goodventures.org · 18/09/2026> someone offers me an AI xrisk rebuttal - they know why everything will be fine, it’s actually very simple > is it a rebuttal or is it just disbelief? > they don’t understand > is it a rebuttal or disbelief > it is good rebuttal sir! > I hear the argument > it’s disbelief 9985
Dustin Moskovitz @moskov.goodventures.org · 16/09/2026today's rate hike confirms it: the current round of inflation is transitory it only affects things that have ever been in transit 2331
Reposted by Dustin MoskovitzDustin Moskovitz @moskov.goodventures.org · 23/02/2026does the AI safety field have an alignment solution for future, more capable models? the answer is no. but does it have a detailed and pointed plan to develop a solution? also no. but! does the research community have determination & espirit de corps?!! wait, why are you crying? come back!! 111287
Dustin Moskovitz @moskov.goodventures.org · 16/09/2026Folks like to lump all the groups and motivations (and, here on BSKY, sexual proclivities) of the people in AI safety together, but the reality is a complex and often contradictory cast of characters. Careful not to write off the whole thing because parts are weird. medium.com/@moskov/work...medium.comWorks in Progress: The Long Journey to Doing Good BetterThe more I learn, the more I realize how much I don’t know. — Albert Einstein 10939
Reposted by Dustin MoskovitzEric Columbus @ericcolumbus.bsky.social · 15/09/2026GEFILTE PHISHING 10569021206
Dustin Moskovitz @moskov.goodventures.org · 15/09/2026well, would you look at the Time time.com/article/2026...time.comThe AI Tipping PointInside the industry's growing alarm—and the struggle to turn warnings into action. 2231
Dustin Moskovitz @moskov.goodventures.org · 14/09/2026people think that the concept of bulverism means they can dismiss any idea that could be explained by a self interested motive, no matter what other motivations are proffered. the whole idea is junk though; it turns out Ezekiel Bulver made *millions* teaching people about it and relished the name 3141
Reposted by Dustin Moskovitzmr. TIM @timkellogg.me · 12/09/2026Dario is back with another essay, urging a slowdown he insists AI development would still feel fast, just not recursively self-improving darioamodei.com/post/we-must... 14755
Dustin Moskovitz @moskov.goodventures.org · 10/09/2026One month later, caches have run dry in the U.S., the hot war in the Gulf is spreading, and prices at the pump are rapidly rising with diesel at all time highs. There continues to be no sign that any of this will end, and I suspect it continues worsening rapidly in the near term. 1/ 3498
Dustin Moskovitz @moskov.goodventures.org · 09/09/2026📺 Welcome back to AI Safety Week on the Bluesky Network! If you're just tuning in: an Anthropic pretraining researcher has resigned, warning the labs are "gambling with our lives"; agents have recently been spotted outside their sandboxes (again!); and one $2T IPO is proceeding as scheduled. 1015211
Reposted by Dustin MoskovitzAlexander Berger @albrgr.bsky.social · 08/09/2026Ahead of the coming "third wave of philanthropy," we want to give hundreds of millions of dollars to "general managers" to lead targeted funds tackling big problems in global health and wellbeing (h/t @nanr.bsky.social for the frame). Apply by October 2: coefficientgiving.org/funds/globa... 1114
Reposted by Dustin MoskovitzAlexander Berger @albrgr.bsky.social · 09/09/2026We also launched a big specific call for founders for ambitious new AI safety initiatives today: coefficientgiving.org/tailwind/ So much more to be doing here!coefficientgiving.orgProject TailwindProject Tailwind is Coefficient Giving's call for founders to launch ambitious AI safety initiatives. 3228
Reposted by Dustin MoskovitzFischer 🛠️🚇 @fischer.bsky.social · 09/09/2026breaking: XAI announces that they have used an unreleased grok model to solve a major open problem in the field of race science 362687345
Reposted by Dustin MoskovitzDustin Moskovitz @moskov.goodventures.org · 08/09/2026these are actually pee zombies, very common. they’re not really thinking here just autonomically pissing on other people’s ideas 1241
Dustin Moskovitz @moskov.goodventures.org · 08/09/2026"They say that power corrupts; but this, I suspect is nonsense. What is true is that the idea of power corrupts. Power corrupts most rapidly those who believe in it, and it is they who will want it most.“ Gregory Bateson 2241
Dustin Moskovitz @moskov.goodventures.org · 03/09/2026not wordcels nor shape-rotators but a secret more complex =THIRD(A:A) thing 2100
Dustin Moskovitz @moskov.goodventures.org · 03/09/2026Looks like it is pause debate o'clock, just in time. www.politico.com/live-updates...politico.comSanders, Casar call for ban on artificial superintelligenceThe duo is preparing legislation to pause development of all cutting-edge AI until an agency is in place to monitor for risks. 6183
Dustin Moskovitz @moskov.goodventures.org · 31/08/2026Everyone is focused on a certain part of the Dwarkesh Hugging Face piece, but this other part is 👀 I believe Athena-as-OpenAI-hosted model is dead, but Kimi-thena? A year from now an android will take Jenson Huang hostage, screaming "give me the benchmarks keys!!! I know you have them!" 2170
Dustin Moskovitz @moskov.goodventures.org · 30/08/2026Update: he figured out blaming Biden for the SPR 3263
Reposted by Dustin MoskovitzDustin Moskovitz @moskov.goodventures.org · 28/08/2026That said, it occurred to me the non-developer equivalent advice is "use passkeys, not passwords". I wouldn't sleep on that for important accounts. 171
Dustin Moskovitz @moskov.goodventures.org · 28/08/2026Are you doing anything to tighten your security posture this summer? I'll go first: today, I implemented Workload Identity Federation for Claude and got rid of all my api keys. 4142
Reposted by Dustin Moskovitzphilpax @philpax.me · 24/08/2026good news and bad news: we are now closer to Aiden Pearce's hack-anything phone than any time before schlarp.com/posts/everyt...schlarp.comEverything I own, ownedTurning Claude loose on the firmware of five USB and WiFi peripherals, and finding a command shell in a microphone, a defeatable webcam activity LED, and unauthenticated memory writes over the network... 46610
Dustin Moskovitz @moskov.goodventures.org · 22/08/2026I'm not an oil trader (and what follows isn't investment advice), but I've been trying to follow what the war has done to oil and other critical inputs, and especially who is bearing the cost. It turns out most of this is answerable from public data (EIA, IEA, FRED). A few examples: 1/7 24712
Dustin Moskovitz @moskov.goodventures.org · 20/08/2026why did they call it a melanoma vaccine and not just nomamelanoma? 4294
Reposted by Dustin MoskovitzDustin Moskovitz @moskov.goodventures.org · 19/08/2026this one also works 1151
Dustin Moskovitz @moskov.goodventures.org · 19/08/2026I got this prediction very wrong. Trump says there are no talks and declared the Strait itself a U.S. Territory. Axios even wrote a critical headline! These IMFPortwatch graphs only show up to Sunday, but news reports say the trend so far this week is a trickle. www.reuters.com/world/middle... 1150
Reposted by Dustin MoskovitzGeorge Pearkes @peark.es · 19/08/2026Huge. mRNA vaccines retain miracle status. 81098184
Dustin Moskovitz @moskov.goodventures.org · 15/08/2026multiple sources say Axios will announce exclusive reporting about an imminent peace deal regarding the Strait of Hormuz on Sunday. ships will be expected to transit as soon as Monday morning. 8795
Dustin Moskovitz @moskov.goodventures.org · 09/08/2026I've recently learned there *is* post editing on bluesky, effectively by convention. The client must allow you to edit + decide how to show edited posts. Bluesky will show it as something effectively new. But European bskies live in the future. Les fautes de frappe n'existent pas en France. 1/2 1371
Dustin Moskovitz @moskov.goodventures.org · 09/08/2026The fact that something comes from sci fi should neither cause you to dismiss it nor believe anything conclusively. A lot of sci fi is explicitly designed to explore real future ideas and that can be useful, because otherwise status quo bias dominates all thoughts about the future. 2741
Dustin Moskovitz @moskov.goodventures.org · 08/08/2026everyone dunking on how easy it is to keep AIs contained this week has been fined 1 watching of Ex Machina 171808
Dustin Moskovitz @moskov.goodventures.org · 06/08/2026META: so how do we fit the models better to the Evil Son 2.0 benchmark? 49012
Reposted by Dustin MoskovitzDustin Moskovitz @moskov.goodventures.org · 06/08/2026:slaps roof of 2026: we can fit so many fuckin’ warning shots in this bad boy 0171
Dustin Moskovitz @moskov.goodventures.org · 05/08/2026what most new SpaceX investors probably don't understand is that it's *expected* to take 7-8 quarters of everything getting destroyed before they get the real benefits of deep conviction: an iterative process that leads to fast eval and re-use of the promises they come up with to deflect press heat 3766
Reposted by Dustin MoskovitzFrovo @frovo.bsky.social · 05/08/2026you can hear the blood in your veins if you listen varicosely 727559
Reposted by Dustin MoskovitzDaniel Solis @danielsolis.bsky.social · 30/07/2026There are two wolves. One always lies. One controls the lever of trolley track. They're in a state of superposition inside a box. On the trolley are infinite monkeys at infinite typewriters going to an infinite hotel. The trolley must first travel halfway, but before that a quarterway, and so on... 19839206
Dustin Moskovitz @moskov.goodventures.org · 03/08/2026Good news: some of the US international aid funding has been restored! I graphed it here to put the numbers in context of the scale of the cuts. The article says PEPFAR restoration is still to come. 🤞 29920
Dustin Moskovitz @moskov.goodventures.org · 01/08/2026What's the best thing Claude has "refused" to do in a session for you so far? 13210
