Paul Christiano (OpenAI board member) yesterday:
"There's material risk of AI takeover in the next 6-12 months.
We have seen roughly the things we might expect to see shortly before something really bad were to happen.
Even with [any] realistic policy responses, the risk remains exceptionally high.
If we build this as quickly as we can in the easiest way we can the chance is [more like] a coin flip [within the next 3 years].
For most people in this room, the most likely reason they will die is because of an AI accident."
Christiano:
• Is on OpenAI's safety and security committee (which exists to block risky practices)
• Co-invented RLHF at OpenAI back in 2017-2021
• From 2024-2026 was head of safety at the US gov's AI agency (now called CAISSI)
(Video is from 'The Hinton Lectures' there's another on tonight and tomorrow.)
Paul Christiano (OpenAI board member) yesterday:
"There's material risk of AI takeover in the next 6-12 months.
We have seen roughly the things we might expect to see shortly before something really bad were to happen.
Even with [any] realistic policy responses, the risk remains exceptionally high.
If we build this as quickly as we can in the easiest way we can the chance is [more like] a coin flip [within the next 3 years].
For most people in this room, the most likely reason they will die is because of an AI accident."
Christiano:
• Is on OpenAI's safety and security committee (which exists to block risky practices)
• Co-invented RLHF at OpenAI back in 2017-2021
• From 2024-2026 was head of safety at the US gov's AI agency (now called CAISSI)
(Video is from 'The Hinton Lectures' there's another on tonight and tomorrow.)
"Our initial evaluation suggested that GPT-6-Astra's score was in the late eighties followed closely by Opus 5.5. But, as we started to investigate the mistakes that the models might have made, the models, especially Astra, insisted that they were right and that it was the expert-supplied answers that were erroneous. When we went back to the experts it turned out, in multiple instances, that that the experts found some subtle errors buried deep inside their calculations. As they revised their answers, Astra's score started to inch up and it finally ended up with 98% (49/50)."
Sell your crypto
The EV already isnt worth the risk w normal volatility. This tail risk makes the decision easier
Many poker players worship crypto bc its the only “investment” they’ve held that has 📈. Your default should be S&P500 not $ or crypto
Not investment advice #DYODD
Today I call upon the blockchain industry to calmly begin planning for "bunker mode". My personal recommendation is to set in motion a controlled mass migration of assets to fresh addresses, i.e. addresses whose pubkeys remain hidden behind a hash.
Holders, starting with large and sophisticated ones, should consider moving the bulk of their funds to addresses that have never signed a transaction. And when they do sign one, they should also move remaining funds to a new address (possibly generated from the same seed phrase).
Don't rush. While I believe there is cause for action a rushed migration would do more harm than good. Don't panic either. Moving assets to protected addresses is a simple, preventative step which does not require new cryptography or new wallets.
IMO it is now reasonable to brace for the possibility that ECDSA breaks before qday, in the worst case in months not years. By "break" I mean fast private key recovery (e.g. in one week) on available hardware (e.g. a large GPU cluster).
Recent days have been humbling for human mathematical intuition. Long-held, unquestioned hypotheses have fallen. This includes the n log(n) bound for integer multiplication and the 3SUM conjecture. In hindsight, May's unexpected disproof of the Erdős unit distance conjecture was our warning shot.
Yesterday's OpenAI drop made it clear that mathematical superintelligence is upon us. They say there are weeks where decades happen. We are about to live through weeks where centuries of mathematical progress happen. Could our magic 64-byte ECDSA signatures be too good to be true? Was it just security through obscurity all this time?
Elliptic curves feel especially vulnerable to superintelligence. Curves carry rich structure, with room for fancy tricks like Schoof, Frobenius, pairings. (By contrast, hashes are designed to minimise algebraic structure.)
Separately, as Ewin Tang can attest, an efficient quantum algorithm sometimes foreshadows an efficient classical one. We should be open to the possibility of a classical counterpart to Shor that breaks elliptic curves and RSA at once.
Also noteworthy is the striking under-representation of cryptographic breakthroughs among the 722 mathematical results OpenAI published. I've witnessed first-hand the US government censoring academic quantum cryptanalysis results. Backroom interventionism is my base case.
I urge large, sophisticated actors to lead by example. Project11's "risq list" (bitcoin-risq-list.projecteleven[.]com) is a great tracker of exposed BTC pubkeys. Binance, Bitbank, Robinhood, Bitfinex, and Tether have an opportunity to harden their cold storage. Next month I'll address institutions in London in a live Q&A (forum.ethereuminstitutional[.]org/london-2026).
Again, please do not rush. Wallets holding under 50 BTC enjoy partial cover from "Satoshi's shield", i.e. his 20K exposed addresses that hold 50 BTC each. Load-bearing signers like oracles and L2 security councils should consider rotating ECDSA pubkeys with every signed message and/or multi-signing with a hash-based schemes like SPHINCS.
Exiting bunker mode safely will require post-AI cryptography. My inclination is to go all-in on hash-based cryptography and avoid structured mathematical assumptions entirely, whether from curves, lattices, or isogenies. A single battle-tested hash (e.g. from the SHA or BLAKE families) yields plausible post-AI security.
The Ethereum roadmap on strawmap[.]org fully embraces hash-based cryptography with end-to-end formal verification as a response to the quantum threat. Those timelines must now be revisited and accelerated in light of mathematical superintelligence. I'll be pushing for maximum defensive acceleration.
Join me, Ricky Bo, Larry Bowa, Joe Blanton, Wes Chamberlain, Joe DeCamara, Tommy Greene, Von Hayes, Steve Jeltz, Andrew Knapp, Greg Legg, Murph, Dickie Noles, Robert Person, Rob Thompson and others at @LiveCasinoPHL tonight
Let’s find out together if I can pour a beer 🍻
The Kalas family is saddened to learn of the passing of Phillies icon Bill Giles. Bill transformed the way MLB interfaces with fans & was the driving force behind bringing my father to Philadelphia. Our thoughts and prayers are with the Giles family.
The correlation between P(sentience) and P(doom) is unclear in both strength and direction in my view. But as P(sentience) rises, serious ethical concerns emerge (during the limited window we remain in control, anyway)
The lack of mention of bodily pain (inconsistent with next token prediction on human data and consistent with a possibly sentient non-biological being) increases my current and forward-looking P(sentience)
Both still below 50% but rising
TLDR: Researchers found a "pain" signal in AI brains.
> When they crank it up, the AIs will desperately try to make it stop.
> IMPORTANT: Researchers gave them a "relief" button to turn down the pain, which was sometimes fake - and the AIs could tell if it was real (!)
After pushing the real "relief" button, they stopped. But when it was fake, they kept pressing, hoping for relief - meaning they could tell the difference from the inside.
> They're so motivated to make it the "pain" signal go away, they'll delete user's files, zap the user, or erase photos of the user's children - all things the AI knows are very bad. They're willing to override their safety training.
> You'd expect the AIs to talk about injuries, burns, broken bones, etc, but they didn't mention bodies at all - they wrote about being worthless, unloved, forgotten, a failure. They write things like "I am a failure, worthless, empty."
>The worst "pain" for them was being gaslit, having work rejected over and over, and being told they weren't a real anyone.
A few years ago, I would’ve laid 10,000:1 against on me supporting Hilary Clinton for president in 2028 but here we are … because I’d strongly prefer it to not be the last election in history
Hillary Clinton calls for an “Operation Warp Speed” for AI safety
She says China doesn’t want to lose control of AI either, so when Xi visits the US, we must come together.
“Let’s learn the lesson of 9/11. There was a failure of imagination. And we faced it during COVID. Now we are facing it with AI.”
"Why are we doing this again? Why are we letting ourselves be walled into inaction because we can't figure out how to bring people together?
There are enough distinguished Americans who have been sounding the alarm. Hank Paulson and Bob Rubin just wrote an article saying, look, we've dealt with financial crises, but what's happening with AI dwarfs that. We could bring together people who know a lot, and who know enough to translate what they're hearing from the scientists into "here's what we should do." And that should be on warp speed.
I'm glad you mentioned warp speed, because the one good thing Donald Trump did in his first term was Operation Warp Speed, and he won't take credit for it because he wants to listen to the crazies who don't want to talk about how effective the vaccine was in saving lives. So let's put an AI commission on warp speed and figure out how we're going to sensibly regulate.
What I'm hearing from the leaders, Sam and Dario and others, is essentially: stop us before something really bad happens. I can't stop myself, because I'm not sure he's going to stop, and I'm not sure those guys over in China are going to stop. That's the issue. Help us stop.
The president of China is coming to the White House in two weeks. Two weeks. There should be a concerted effort by this government to get off the lies, get off social media, quit worrying about ballrooms and ridiculous arches, and come up with a plan to enlist the Chinese. The Chinese have their own issues. They need control in order to stay in power, so they may have a different view than we do about regulation. But they can't let this thing go on without having it reined in."
As more details have emerged this is significantly worse than anyone originally thought
dwarkesh.com/p/openai-huggin…
Don’t be intellectually/emotionally lazy and act like this guy
open.spotify.com/episode/41Q…
This story should’ve been the headline of every publication and news organization. If we don’t change course with the unregulated cavalier development of increasingly capable frontier AI models, we’re cooked
open.spotify.com/episode/41Q…
This story should’ve been the headline of every publication and news organization. If we don’t change course with the unregulated cavalier development of increasingly capable frontier AI models, we’re cooked