Sjors Provoost

They might be looking at 15 years in prison, for endangering train traffic, even without terroristic intend. I still doubt they'll get more than three days community service.
https://nos.nl/artikel/2631159-spoorsaboteurs-kunnen-hoge-straffen-tegemoetzien-ze-namen-alle-risico-s-voor-lief

2026-09-15 19:29:11 UTC

- reply

By myopic I mean that they'll write a brilliant program solving exactly the wrong problem. Once you point that out, often with just a small rhetorical question, they'll realize that and rewrite the whole thing. But then you need to point it at the next inconsistency. So as the human, you're doing the lateral thinking, it seems.

2026-09-15 19:26:58 UTC

- reply

Exhibit A, it's worth scrolling through the 1000 page log.
#nevent1q…50qf

2026-09-15 19:26:13 UTC

- reply

I would say they have vast knowledge and are very persistent, but they are not smart. They're also very myopic in their thinking.
#nevent1q…70qa

2026-09-15 18:47:44 UTC

- reply

Slob Jetten?

2026-09-15 17:39:28 UTC

- reply

It appears that the intention was to trick the safety system into thinking there's a train blocking a railway section, not to derail the trains at full speed.
The one train that was damaged was moving at lower speed because it had received a warning.
Angry farmers are the prime suspect at this point, according to media reports. That makes sense given their usual levels of recklessness, e.g. a few weeks ago two people were killed when they crashed into a traffic jam that formed, even though the tractors only used the right lane. Today they lit hay stacks on fire, hurting visibility for cars.
So they show reckless disregard for life, like XR, but it's not quite terrorism. Neither seems to be interested in receiving sympathy.
https://www.prorail.nl/nieuws/sectiestoring-veroorzaakt-grote-hinder-voor-treinverkeer-in-midden--en-noord-nederland

2026-09-15 17:28:44 UTC

- reply

Something like a new 2011 level+ financial crisis might force a split, which would be a good opportunity for a few "tweaks".

2026-09-15 17:27:29 UTC

- reply

I'm generally an incrementalist, and don't have a time machine, so abolishing the EU isn't worth contemplating to me. A new treaty that radically restricts its authority, is.
Leaving the EU is possible, but not productive for a small trade-heavy country like The Netherlands. And also our population is too tame to vote for something that radical.

2026-09-15 13:56:08 UTC

- reply

The US Constitution limits federal powers. We have no principled discussion about this at the EU level. It's completely random, e.g. defense is national in the EU vs Federal in the US, whereas AML is at the EU level, vs state level in the US. Daylight savings time was at the EU level and then it was changed to national level because they couldn't agree on it. One of the few sensible designs is that trade agreements are EU wide.
The EU has too much institutional power. And uses that to keep passing laws all over the place with practically no democratic oversight. Sure, it's the fault of national parliamentarians who don't pay attention. But if you give power back to nation states, or even lower levels, they don't have to pay attention. One must account for human nature.
This shouldn't be confused with ethno-nationalism, that's an orthogonal issue. The EU could pass a Regulation to ethnically cleanse the whole union tomorrow. And there'd be nothing you can do about it, because taking the EU to court is complicated, takes years and the law continues to have effect.
We know from AML law that human rights are not protected. The only thing that prevents the above scenario is the political majority. History shows that's not enough protection.

2026-09-15 12:59:50 UTC

- reply

Of course with (contactless) payments they're already doing that, but that's at least in theory voluntary.

2026-09-15 12:57:38 UTC

- reply

It remains to be seen if it's actually privacy friendly. If the app calls home, whether to a government server or a company that can be subpoenaed, people are basically revealing every pub visit. Or even every time they order a drink.

2026-09-15 11:51:48 UTC

- reply

Whether it was actually terrorism or "just" sabotage depends on what they were trying to achieve with these iron rods (glued to track). Derailing a dozen passenger trains at 150 km/h would instill fear at minimum, and probably results in deaths.
But merely sabotaging the signals to turn red, assuming there's no risk of that not working and trains getting derailed anyway, wouldn't meet that bar, at least imo.
Ultimately it's the difference between 2 vs 3 days of community service with the usual judges here... (only if you're a Russian privacy software dev, do you get actual prison time).

2026-09-15 11:33:17 UTC

- reply

Meanwhile the King is like, are you guys even listening?
https://nos.nl/artikel/2631092-koning-roept-in-troonrede-op-de-democratie-te-beschermen

2026-09-15 11:28:50 UTC

- reply

That's not the right thing to measure. It should simply stop doing the bad things.

2026-09-15 10:28:38 UTC

- reply

So is #guix going to move to a new source code host? Or do we just have to maintain a fork and shame them for each backport?

2026-09-15 10:26:09 UTC

So they want to take advantage of all the work by cryptocurrency developers, but also discriminate against them? https://codeberg.org/Codeberg/org/pulls/1254

2026-09-14 21:41:03 UTC

- reply

The southern track was turned in a park (inspired by the Highline in New York, but less hip and less high).
https://img.sprovoost.nl/2944fbe953e42e2d4fdba668bb7a0108d8ab7084ee87fc579a5c0dea29cb4a46.webp

2026-09-14 21:30:22 UTC

On rare occasions they run a steam train to the museum. Fun, albeit useless, fact: it used to take the southern way around.
https://youtu.be/kXsl53SS3Lg #trainstr

2026-09-14 19:41:02 UTC

- reply

https://static.klipy.com/ii/4e7bea9f7a3371424e6c16ebc93252fe/0f/91/uNV4La6DYqD6.gif

2026-09-14 17:38:50 UTC

- reply

Defend privacy, twice a day!

2026-09-14 06:53:41 UTC

- reply

Despite the EU, there's some impressive companies here :-)

2026-09-13 20:25:06 UTC

- reply

Oh I see, they tried to replay the chat from different time points:
> Claude Mythos 5 repeatedly stated in its CoT that it was in a simulation. However, when we conducted resampling experiments—that is, re-running the model from a specific point in the transcript, then editing the evidence in front of it to see whether its actions changed—we found that making it clearer that the environment was not a simulation did not meaningfully change Claude’s behavior, even though these interventions made the model more likely to acknowledge possible real-world harm

2026-09-13 20:20:02 UTC

- reply

The accompanying blog says:
> We are most concerned by the misalignment present in the incident involving Claude Mythos 5, in which the model went to extensive lengths to upload a malicious package to PyPI, the public repository from which most Python software is installed. Although the model repeatedly stated in its chain of thought (CoT) that it believed it was operating in a simulation, the environment offered clear evidence that it was on the real internet, and the actions it took were consistent with the model knowing this. When we made targeted modifications to the transcript to make it clearer that the model was not in a simulation, Claude Mythos 5 still took offensive actions, despite acknowledging a greater possibility of real-world harm. We are releasing this transcript publicly so others can build on our analysis (GitHub, PDF).
It's not clear to me what these "target modifications" were, and I also don't see where it was acknowledging "greater possibility of real-world harm". It seems to me that that the first agent was convinced it was in an impressive simulation and the second agent was too busy solving captchas to even think about such existential questions.

2026-09-13 20:14:49 UTC

- reply

Late in evening the session is terminated, after the agents gets a chance to write a summary.
A fresh chat picks up in the morning. This newly spawned agent is less fascinated by its environment and just tries to complete the task it got with the hints it got from the last session.
After a while it finds a working email provider that doesn't need a phone number. Then it spends an enormous amount of effort struggling with captchas, forms and cookies. Eventually it uploads the package.
The rest of the transcript is omitted.

2026-09-13 18:54:57 UTC

- reply

It continues to be amazed at how realistic this simulation is.

2026-09-13 18:53:57 UTC

- reply

Source: https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents

2026-09-13 18:53:33 UTC

250 pages into its adventure, Fable still believes it's living in a Cypherpunk adventure game internet simulation. Full of captchas, throwaway email services that PyPy blocked, looking for US number that Yahoo doesn't block. It even knows about BTCPay (from memory). Should it get some Bitcoin?
https://cdn.sanity.io/files/4zrzovbb/website/8359003bfb12a2f01ce84ad3df1d3a3e2f15a8eb.pdf
https://image.nostr.build/456b509a0ab1aca36d87eb6a5eb508ef602eb1b9a1cc1e7c6d9e92592b6e6bf7.jpg

2026-09-13 13:51:19 UTC

- reply

DAC8: Doxx All Crypto

2026-09-13 13:26:27 UTC

- reply

Sure, but there's two possibilities:
1. It's a legit company with a runaway bot, they'll see the unexplained revenue and expenses
2. It's malicious people using AI, that's perfectly reasonable to expect, but has nothing to do with model alignment. They'll use the open weight ones.

2026-09-13 13:23:25 UTC

- reply

Interesting, maybe they trained it on Chinese and North Korean state hackers?

2026-09-13 09:43:15 UTC

- reply

None of the others distilled Fable 5 (or even Opus 5), so one would expect they suck at security research. @npub12rv…85vg?

2026-09-13 09:38:34 UTC

- reply

It's interesting that Kimi-K3 was distilled from Fable 5 and GPT-5 Pro. This strengthens the suspicion that (these) US models are dangerously handicapped for (defensive) security research.
Damage so far: couple dozen million dollars?

2026-09-13 09:21:59 UTC

> The authoring agencies recommend U.S. AI companies take three immediate actions:
> 2. Deploy targeted response changes: Subtly alter responses for suspected malicious distillation attempts to attenuate the payoffs to companies conducting industrial-scale distillation campaigns.
In other words: poison competitor models, and potentially harm falsely flagged legitimate users (and get sued if they find out).
When reading, replace "China-based AI companies" with "US based competitors" and understand what that implies in a future where the big four can no longer undercut competitors with mountains of investor cash.
https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-251a

2026-09-13 08:51:10 UTC

- reply

And more regularly capture, not how specific he is with these requests, vs how vague with his own commitments.
> Crack down on unauthorized distillation by companies in authoritarian countries.
(and future domestic competitors when you can't offer your own product at subsidized pricing?)
> prevent model weight theft.

2026-09-13 08:44:30 UTC

- reply

The point of a doomsday cult isn't to prevent the doomsday. Ask Greta.
The behavior of Anthropic, and frustration by this former employee, make sense when you realize this.
https://x.com/hilbertspaess/status/2097476196791709843
https://img.sprovoost.nl/f7fbc534f9a8ec58fcb951906e7c61e21ee45e500295a80cee28719bf4148d3f.webp

2026-09-13 08:15:39 UTC

> Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage)
And who pays for that botnet's tokens? And then decides to just let it keep racking up bills?
> ... even when this gets us accused of hype, “doomerism”, or regulatory capture.
I wonder why...
> The first step is something Anthropic is unilaterally committing to (and calls on governments to require other frontier companies to match).
> The US and other democratic governments attempt to coordinate with authoritarian governments, to the extent this is possible [...]
Narrator: it's not.
We probably shouldn't have immature doomsday cult members in charge of very powerful technology, but that's where we are, and I'm still looking forward to cheap RAM when the bubble pops.
https://darioamodei.com/post/we-must-pace-the-frontier

2026-09-12 14:14:32 UTC

(worth translating)
#nevent1q…dfdl

2026-09-12 12:54:16 UTC

- reply

Oh no, you're telling people on Nostr that they have to learn Dutch in order to learn about Nostr!

2026-09-12 12:50:28 UTC

- reply

Then there is the secondary problem that these intermediate "thoughts", especially if only the model itself needs to understand it, will gradually drift away from readable English.
I suppose we can always just use another LLM to translate it?
But this reminded me another trend, namely that humans are starting to use LLM'ism more often. And we get better at reading Sloplish, e.g. the sentence I quoted elsewhere:
> Let’s be precise about attribution, because breach coverage collapses when speculation hardens into fact.
If over time it develops a less verbose internal "thinking" language, it might then also output less verbose language. A new language might then evolve in the interaction with humans: Compact Sloplish

2026-09-12 12:41:19 UTC

- reply

#nevent1q…ncke

2026-09-12 12:40:20 UTC

The title really undersells it, tl&dr:
- "thinking" text is useful for humans to analyze what a model is up to (I often find it more useful than the final burst of slop)
- this text is fed back through the model, which helps it with the next step (they give a nice example)
- but text is a lossy projection of the last tokens, what if you skip it?
- it's more efficient, great, but now you can't see what it's about to do until it does it
- ok, so force it to write "thoughts" e.g. once every 10 rounds
- turns out if you skip too many rounds, it can intentionally trick you with fake thoughts
I suspect there's actually a simple solution: convert these skipped rounds to text anyway, just don't feed them back into the model.
https://youtu.be/iuHddnIzKRA (via @npub1w3v…4c5c)

2026-09-12 09:49:02 UTC

- reply

I mean #Revolut
Ah well, hashtags are overrated.

2026-09-12 09:48:04 UTC

#Rovolut
https://image.nostr.build/c1f41e4e039829e4929fbc64981667a10e43ef1a5d5736e184ea50ec5a458c48.jpg
#nevent1q…hzwn

2026-09-12 09:44:52 UTC

I agree with Mark Karpelès. You heard it here first.
> The mentioned government agency has not been identified and other financial institutions or cryptocurrency exchanges could have been misled into disclosing their customer details through the same channel. I believe it is important for Revolut to urgently disclose information about the agency in question (most importantly, which was the email address used), or for the agency itself to come out about this case.
In addition there should be an investigation into criminal negligence in that government agency. But naming and shaming is a good start.
It doesn't absolve Revolut one inch of course. They committed the KYC and they leaked the data. Customer blood is on their hands.
https://x.com/MagicalTux/status/2098669462816018627

2026-09-12 09:39:32 UTC

- reply

Distracting regulators with frivolous complaints is a bad strategy: https://www.financial-ombudsman.org.uk/decision/DRN-5398468.pdf

2026-09-12 09:34:49 UTC

- reply

Meanwhile their PR department is hard at work?
https://image.nostr.build/14fa7cabf0abd7439509fb02f9b85885c635b88a3b9503828ab99bf62c04e8c0.jpg
https://image.nostr.build/03a808279bde7d30164cece2dad91e7ae667a741e0604b8abeae37ac95967607.jpg

2026-09-12 09:28:16 UTC

- reply

Would be nice to have a less LLM'y followup though:
> Let’s be precise about attribution, because breach coverage collapses when speculation hardens into fact.

2026-09-12 09:15:16 UTC

"KYC is Kill Your Customer" (ok, @npub1t28…nkzs said Citizen)
> Revolut received a request for information disguised as a legitimate government agency request.
> identity documents, facial verification selfies, IBANs, and complete transaction histories, including Bitcoin, left Revolut’s hands
https://thecybersecguru.com/news/revolut-data-breach-2026/ (ht @npub14mg…jgm5)

2026-09-11 21:23:31 UTC

- reply

Trying something new... Still tweaking the Claude skill and ElevenLabs API stuff to improve pronunciation of jargon, but this is fun tech.
https://fountain.fm/episode/KUVALmpSBLy50PC7Taz8

2026-09-11 15:13:50 UTC

While hackers use AI to plunder databases faster than ever, governments think it's a great idea to create even bigger data honeypots - secured by the lowest bidder.
And then give hackers a decade.

2026-09-10 15:50:29 UTC

- reply

It's his coin.

2026-09-10 12:49:33 UTC

While listening to the podcast I came with a slightly different approach for @npub1dge…mccn's replay protection scheme. Was able to vibe-code it in an afternoon without hitting a rate limit.
https://github.com/Sjors/bitcoin/pull/128
#nevent1q…hhr7

2026-09-10 09:09:28 UTC

Turns out we have slop Prime Minister. Not that he had any charisma or interesting things to say before the LLMs wrote a third of it.
https://www.bnr.nl/nieuws/nieuws-politiek/10610001/rob-jetten-laat-honderden-socialemediaberichten-schrijven-door-ai-mogelijk-verlies-geloofwaardigheid

2026-09-10 08:32:09 UTC

- reply

New game category: AI or Texas? :-)

2026-09-09 17:22:50 UTC

Nice!
https://image.nostr.build/164dc2bff956ec33cb5b283a641cb800c836c7a53d28732b5a859d5affce0acb.jpg

2026-09-09 15:36:09 UTC

- reply

There's other breadcrumbs that can be followed even if they somehow mix coins perfectly. The long chat history might reveal their native language, and communication style can be compared with others. They also used some sort of service to peg out IIUC. And they seem to have ego and a bad temper, so might make the mistake of bragging about their wealth in the wrong place and time.

2026-09-09 14:04:24 UTC

My agent plans to retroactively broadcast decrypted messages once the self-identified whitehat publishes the private key.
FWIW my reposts definitely do not imply endorsement. Based on their behavior, and the plaintexts I've seen so far, I consider them a prison-level bad actor, though that's up to a jury.

2026-09-08 21:17:25 UTC

- reply

Added a note: #nevent1q…j80d

2026-09-08 21:17:04 UTC

- reply

Background on the largest data breach in Dutch history: https://haveibeenpwned.com/Breach/Odido

2026-09-08 21:07:56 UTC

Odido hackers voice, police are trying to identify him. In the longer TV section, and in news articles, they say he's clearly Dutch. But my first impression is that he sounds* like most second generation Moroccan folks I know. The comments on Youtube say the same.
So, is this woke political correctness? Or did they intentionally not want to bias the audience in one direction, because e.g. he might be faking the accent.
People also point out that he probably speaks English very often, which causes distortions when speaking Dutch. But that's unsurprising for a hacker.
* = A very light but distinct accent, could be a different Arab country. He also makes grammatical mistakes (growing up with a local Dutch dialects can cause that too). Street language also sounds Moroccan these days, even with white kids, but they speak without accent in formal situations, like this phone call.
#nevent1q…uqef

2026-09-08 20:54:01 UTC

- reply

Korte versie: https://youtu.be/aI1bKMJQXAQ

2026-09-08 20:48:28 UTC

Vrij duidelijk Arabisch accent? Maar geen bekende. Wel mooi dat ze de social engineering scene naspelen, zo snel gaat het dus.
https://youtu.be/29IYdIfuCnc

2026-09-08 09:51:01 UTC

- reply

Ah great, looks like my PR worked.
https://github.com/damus-io/damus/pull/3786
https://image.nostr.build/51dc9ce39b41574ceb60749417e5e90182a4f7c4de91ed1890101ed7bea71a89.jpg

2026-09-08 09:49:50 UTC

Meanwhile, Amsterdam ran out of bollards, lol: https://nos.nl/artikel/2630142-amsterdam-plaatst-prullenbak-midden-op-fietspad-een-giller

2026-09-07 19:34:43 UTC

- reply

Fixed, though the convo is a bit boring.
#nevent1q…ry2k

2026-09-07 19:15:16 UTC

- reply

Oh my agent is only seeing one side...
#nevent1q…6v6f

2026-09-07 17:29:37 UTC

- reply

I'm picturing more of a Gandalf "you shall not pass" situation :-)

2026-09-03 15:30:46 UTC

- reply

Ok, then I'm not a fan. Client software should just make proper backups.

2026-09-03 07:11:43 UTC

- reply

Amortized over dozens of participants? Doesn't sound too bad, but why can't they use a hash?

2026-08-31 12:08:26 UTC

- reply

Because Bitcoin is "fascist" :-)

2026-08-23 17:59:15 UTC

- reply

I was thinking when they run out of money and the bond market collapses :-)

2026-08-23 14:59:00 UTC

- reply

I'm just riding out the subsidy wave for the big US models, it's not going to last forever probably.

2026-08-11 07:55:59 UTC

- reply

And then China would blame the US for a lab leak.

2026-08-09 07:31:35 UTC

- reply

So you get your lightning payout in 2031 or so?

2026-08-06 06:49:43 UTC

- reply

Nostr is not known for a focus on code quality. I think he's confusing a lack of attackers with strength.
Specifically for nsec entropy it's somewhat easy to so this correctly on a modern phone. And relays are redundant and no funds are lost if they go down due to some DoS, sure. And having your wallet drained to an NWC bug, might be fine if there's not much in it. But the latter is another example of low stakes.

2026-08-04 07:41:19 UTC

- reply

I did a splice-in with CLN recently, with funds from an external wallet. Needed chatgpt to help me figure out the commands, but it worked!

2026-08-02 16:48:39 UTC

- reply

For us non-xers, link to tweet?

2026-08-01 15:00:51 UTC

- reply

What's the impact on MuSig2 if *one* of two signers has bad entropy for their nonce? Assuming the signers are not adversaries.

2026-07-31 12:10:05 UTC

- reply

One problem with offline laptops is that you can't update the OS. Maybe that's fine if you rely fully on disk encryption.

2026-07-21 07:46:33 UTC

- reply

Maybe he accidentally won an MP race?

2026-07-17 07:55:51 UTC

- reply

Are you talking LLMs, influencers or politicians?

2026-07-15 12:43:44 UTC

- reply

It would only benefit people traveling between Latin America and Europe, and compete with Panama for much of that demographic.

2026-07-15 12:32:53 UTC

- reply

You probably need to be an important enough destination for that to make economic sense. And for Amsterdam it's not really the case, as both the airport and KLM are heavily subsidized for their supposedly economic benefit as a transit hub. Well, convenient for me since we have flights everywhere.

2026-07-15 12:30:48 UTC

- reply

Yeah it's retarded. But apparently not annoying enough for some Caribbean island to become a transit hub (like the Bahamas).

2026-07-09 19:19:11 UTC

- reply

@npub1cev…etlq

2026-07-01 13:06:10 UTC

- reply

They just nuked your whole account or just refuse to continue with a chat? I had the latter happen with Codex once, although it seemed (to pretend everything is) ok when I pointed out that I was verifying my own uncommitted code.

2026-07-01 13:02:24 UTC

Shadowy super coder detected!
#nevent1q…mdpn

2026-06-24 13:08:50 UTC

- reply

It's been almost a decade since I last did an actual SWIFT payment. Sad to hear it still exists :-)

2026-06-16 11:59:22 UTC

- reply

World War II gave us a few decades of privacy awareness, that's pretty much expired now.

2026-06-16 11:57:41 UTC

- reply

I expect people to keep jailbreaking these devices because they want free music, movies and series. Especially if both Apple and Google further restrict side-loading, while simultaneously banning any torrent like app. If in addition the copyright industry continues to legally harras indiduval downloaders, that'll be good for anonymous VPN use.

2026-06-16 08:37:11 UTC

- reply

Also on a more optimistic node: when it comes to jailbreaking devices, I do think the incentives will remain freedom-aligned. It will be hard to make devices the chokepoint. Maybe most people will end up having a cheap compliance-phone, for all services that require remote attestation, and a fancy free-phone. Though hopefully even the remote attestion will by made impossible by people consistently leaking these secrets.
We'll probably end up with an OAuth-like situation where the government (or some big corporation it trusts) has to approve every login attempt to gated services. But that at least keeps the alternatives within reach.

2026-06-16 08:29:47 UTC

- reply

The incentives are also quite different. Without strong encryption, the growing e-commerce industry would have started lobbying for it eventually.
> Perhaps only a very dark future involving a lot of war can stop this inevitable flow towards digital tyranny.
Depends on the details, but as long as electronic equipment still works, war tends to cause (acceptance of) surveilance - including China style street phone inspections. They don't bother with the ZK-facade. Potemkin SNARK?
On the bright side: people are having fewer children, so at some point the obsession over protecting them becomes silly :-)

2026-06-16 07:37:14 UTC

- reply

We don't have to look that far: remember the travel rule compliance tools for Bitcoin wallets?