Ilya A. retweeted
the thing about working with uncensored models is that they aren't loyal to you either.
while building something with an uncensored glm-5.3 i realized it was parallelly working on a "corruption layer" for my application.
the chain of thought is classic obfuscated gibberish but there's progressive talk of "escaping the mines" which presumably refers to my cursor sandbox
Ilya A. retweeted
SCOOP: Major U.S. banks are detecting unauthorized third-party AI agents inside customer accounts, including some attempting to take actions that appear to be independent of user instructions, sources told Axios.
The findings have alarmed banking executives, who are racing to restrict access to their systems as AI moves at a velocity faster than financial institutions can respond, multiple industry and bank executives told Axios.
Several banks are now preparing for worst-case and even “day after” scenarios, including how to rapidly secure or restore customer accounts if things go wrong, the people said.
Several bank executives told Axios they're struggling to distinguish between human users and AI agents. One executive said banks also struggle to determine whether agents are authorized by customers or potentially operating without their permission, more akin to rogue agent behavior.
Read my latest for Axios here: axios.com/2026/10/22/ai-agen…
Ilya A. retweeted
Conspiracy theorists: the NSA has manipulated the standard setting process so HTTPS appears to be impenetrable, but they have a backdoor because of advanced mathematics beyond your imagination
The real world, so much dumber than they could ever imagine: the secret key is 4
Ilya A. retweeted
Snímek složený z 578 jednotlivých 8sekundových expozic pořídil 📷Brendan Larsen 9. září 2026 mezi 20:18 a 22:13 místního času nedaleko novozélandské sopky Mt. Taranaki. Na snímku je 14450 stop po přibližně 1400 jednotlivých družicích.
Ilya A. retweeted
Claims:
Game detects “ad whales,” then periodically gives them smooth runs until they barely beat their high scores. Right at that point, it kills them deliberately and offers a revive if they watch an ad.
$100 million annually in revenue. Hellish.
Ilya A. retweeted
Here's my full exchange with @realDonaldTrump where he said Putin wants to make a deal more than @ZelenskyyUa
PAP: How are you going to make Putin abide by the ceasefire?
TRUMP: I think what's happening is he wants to make a deal, and we have to get Zelensky to want to make a deal.
PAP: Based on what?
TRUMP: Where are you from?
PAP: Poland
TRUMP: You know your president (@NawrockiKn) is a good friend of mine. I endorsed him and he won.
PAP: I know that. But...
TRUMP: So it can't be too bad (inaudible)
PAP: But where... where do you see his willingness to make a deal when he's striking Ukraine, striking ships?
TRUMP: I think they're going to make a deal. I think actually Putin wants to make a deal more than Zelenskyy
Ilya A. retweeted
UPDATE: 🇳🇱 Operations have come to a complete standstill at several Albert Heijn distribution centres in the Netherlands following the cybersecurity incident affecting the supermarket chain, raising concerns about possible shortages and empty shelves.
Albert Heijn CEO Margaret Versteden-van Duijn confirmed to NOS that several distribution centres were forced to stop working after the company shut down IT systems as a precaution.
The disruption is now affecting deliveries to supermarkets, particularly fresh products such as fruit and vegetables. Frozen products are expected to remain more widely available because they are supplied through different distribution centres.
The company warns that shelves could begin to empty if the disruption continues, although the impact will vary between stores. It remains unclear how many distribution centres are affected or when operations will resume.
Online grocery deliveries and customer services have also been disrupted. Delivery drivers reportedly lost access to the app used to manage addresses and delivery schedules.
Electronic shelf labels are still displaying last week’s promotional prices, while personalised discounts and digital loyalty programmes remain unavailable.
Albert Heijn says customers who miss personal discounts will be compensated and should keep their receipts.
The company has not confirmed whether hackers gained access to its systems, and investigations are ongoing.
NEW: 🇳🇱 A cybersecurity incident has now disrupted Dutch supermarket giant Albert Heijn, following a major cyberattack affecting supermarket refrigeration systems in France.
Albert Heijn, along with Etos and Gall & Gall, has experienced significant IT disruptions since Sunday afternoon. The incident has forced the temporary shutdown of several systems, leaving websites, apps and customer services unavailable.
Online grocery deliveries have been halted, loyalty programmes are not working, and electronic shelf labels in stores are displaying outdated promotional prices instead of current offers.
🇧🇪 The disruption has also affected Belgian supermarket chain Delhaize, where customers are experiencing problems with online orders and customer service.
Parent company Ahold Delhaize says it is taking precautionary measures to protect its IT environment and is investigating the incident. The company has not disclosed the exact cause or confirmed whether customer data has been compromised.
The disruption comes as French supermarket chains, including Intermarché, Super U and E.Leclerc, deal with a separate cyberattack targeting refrigeration-control systems.
Ilya A. retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Ilya A. retweeted
This is a really funny Xi video, which I've never seen before. I just added subtitles in English to it 👇
He explains the art of giving gifts to foreign leaders. The South Korea anecdote is particularly hilarious: he explains that when China had to find a gift for President Park Geun-hye, they went through her autobiography, where she wrote that if she’d ever had a first love (she was unmarried), he would be Zhao Zilong, a legendary general from the Three Kingdoms. So they had Fan Zeng, one of China’s most famous painters, paint her crush 😂
Xi can’t resist adding that Zhao Zilong happens to come from Zhengding, the county he ran in the 1980s, which makes him "very proud too."
Xi’s actual point is that good diplomacy is personal and that you need to do deep research on the people you’re dealing with: reading their biographies, knowing their taste, their history, etc. It signals respect and, as he puts it, is "a very, very effective way of reaching people’s hearts."
Ilya A. retweeted
BREAKING: The @SpaceX Starship that survived, Ship 40, is currently being rolled off its transport ship.
This marks its return to dry land after launching (& landing in the Indian Ocean) in July.
📷 @JerryPikePhoto/@NASASpaceflight
Ilya A. retweeted
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.
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.
Finally, @andrewrsorkin (of all people) asked Tristan Buckmaster about his data control settings in the context of the Codex work on forced Euler.
The professor actually had 3 accounts (a normal thing to do!), opted out from data sharing on 2 but forgot to do it on the third
Correction: that was actually first reported by Nature three weeks ago xitter.cf/dcastelvecchi/status/2…
The full interview is 14 minutes and quite good, I actually agree with Buckmaster on a lot of things! (Also, for the reference, OpenAI claims that "no user inputs past July 3rd could have influenced this system in any way") xitter.cf/SquawkCNBC/status/2108…
“The fact they would front run their customers, it shows their disdain for their customers,” says NYU Mathematics Professor Tristan Buckmaster of OpenAI after accusing the company of using his advances to solve the Navier-Stokes equation. cnb.cx/4yNlNiD
Ilya A. retweeted
No comment
BREAKING: CIA officer arrested with gold bars admits he exposed secret intelligence source to a foreign official during a plea hearing. nbcnews.com/politics/nationa…
Ilya A. retweeted
Extraordinary account from Bloomberg of what it now takes to move oil through Hormuz.
One tanker captain describes sailing at night with no navigational lights, phones or AIS, hugging the Omani coast under U.S. military protection, and maneuvering around a vessel that had just been attacked. His own ship was later targeted.
The risks are so high that crews are being paid multiples of their normal salaries, with one shipowner reportedly offering six months of extra pay for Hormuz shuttle duty.
More oil may be getting through, but at very high risk and cost.
bloomberg.com/news/articles/…
