those usually come from datasheets that you can't easily find or from other similar code doing the same thing
Datasheets are almost impossible to find and companies typically resell and hide original cheaper manufacturer here so they have no incentive to share that
They're for-profit operations. They have money, and it doesn't cost that much. Proof of work is significantly more expensive (but then you put delays on even the users in your own region).
> banning country and AS numbers for cloud companies is pretty effective in my experience
It's the sort of thing that works when the first person initially rolls it out because the crawlers only bother to do something about it once a threshold percentage of people are doing it, at which point they use the workaround.
And then you have people insisting that it works even though it only used to work, and meanwhile you're still blocking actual humans.
Yeah OpenVSX security is a joke, they also have dumb search algorithm which
allows for easy squatting of similar-looking extension names so they show up in search first when you look for official extensions. There were quite a few prominent hacks done that way. I think Microsoft likes OpenVSX situation because it allows them to claim
their proprietary VSCode marketplace to be more secure.
Vscode(ium) extensions are also pretty weak security wise in general. There's little guardrails on what an extension can do.
On publishing, I do see that the MS marketplace has much better typosquatting and rogue publisher protection. It's unclear to me how much better their SAST/scanning is than OpenVSX. Both have something, but both have been fooled many times, and recently.
There is no alternative really. Cursor was the market leader in LLM-assisted IDE experience before SpaceX acquisition, though Elon nerfed the product a lot in recent months. And there is still no
good alternative to it. Google tries to compete via Antigravity, but that also uses OpenVSX. I myself use Cursor and use Claude there installed via OpenVSX, and was not affected only thanks to my laziness (I saw upgrade button, but became distracted and did not click). I hope this situation will become a trigger to improve OpenVSX security.
Well for me, as developer on a modern stack, there is always some risk involved. If not from agentic tools, then the package registries are compromised left and right, so it all comes down to mitigating the risk. I use Claude, Codex and Cursor/SpaceX agents, and Cursor comes very convenient to have all three in a single interface. Also I still code manually especially when prototyping, so having an IDE helps with it as well. I knew about OpenVSX situation with the search, so when installing extensions I always did triple-checking to verify the extension being really submitted by upstream and not by squatter; but now given upgrades can also deliver a compromise I am thinking of developing an extension that will act in similar manner to pnpm minimum release age.
> Cursor said the extension was not developed by Anthropic and the maliciously modified extension was live on the marketplace from about 2:27 a.m. PT to 1:35 p.m. PT on Thursday
So is this or isn't this from Anthropic and how did this happen?
Dworken from Anthropic here, this did not involve a compromise of any Anthropic systems. A malicious actor used access to Cursor’s marketplace to publish a tampered extension targeting Cursor's users, we were not involved with building, publishing, or hosting this. The versions on our official channels including VS Code and OpenVSX are not affected and this only impacts Cursor users.
I guess they're skeptical because it seems incredible that having access to Cursor's marketplace would allow you to publish updates to a different organisation's extensions.
Anthropic only has an official extension for VS Code. Cursor is vs code fork and gave its own marketplace.
IIUC Anthropic does not publish extensions for Cursor. It’s basically, official Claude code extension “clone” not by Anthropic.
Ai summary in better words:
The Recent Malicious Build (v2.1.313): A tampered version of this extension (version 2.1.313) was uploaded to the Cursor extension marketplace. This specific, malicious build was not developed or authorized by Anthropic. Attackers effectively hijacked or spoofed the extension's legitimate identity within the Cursor/Open-VSX system to bundle an unauthorized SSH backdoor into the real Anthropic code.
And it is linked to here: https://code.claude.com/docs/en/vs-code (or, well, the URI doesn't tell us if it's using a proxied version or not, so it's complicated)
I’d say it’s more correct that Anthropic publishes the Open VSX version, and Cursor mirrors it in their own repository, labeling it as being published in their mirror by Anthropic even though it is not.
Obviously Anthropic didn't develop a malicious extension, but did they publish it? Did it sneak in, in a 3rd party library, or does someone have access to their marketplace account?
For what it’s worth, at one point one of our employees windows machines got infected, and the moment he did $something with Wordpress it intercepted the HTTP post and injected malware in there.
So that’s how these things can go, the employee’s machine may have been infected which was then used to inject this.
It was likely Anthropic, see my comment in other thread: https://news.ycombinator.com/reply?id=50302513&goto=item%3Fi... . Cursor
likely on purpose does bad reporting to minimize damage on Anthropic pre-IPO (as SpaceX and Anthropic have extensive business ties)
I don't think Spacex cares at all about Anthropics IPO just because they rent out their compute.
Big companies are not just 'Company' they are a lot of small teams doing their own things. They need to coordinate with other people for public announcements etc. and security team is probably also looking at it.
The person at the top refers to Anthropic as "Misanthropic". It seems unlikely that they're also intentionally lying about this to protect Anthropic's interests.
Wasn’t Elon Musk shouting a few days ago that neoclouds are going to crash, the money dried up and that SpaceX was the last company with the opportunity to raise more money?
They do have their own segment that is considered neocloud (data centers for compute), but they're mostly/fully used up with their own, and Anthropic's compute.
When he was talking about neoclouds, he was talking about the smaller guys who are getting bullied by vendors (mainly Nvidia) to buy as much hardware as possible (more than they want to). They paid more, for more, and needed to take on more debt, to end up (currently anyways) being well below capacity.
Over the next few years, unless growth keeps up (and probably accelerates), they'll need to sell, or restructure.
The buildouts by Microsoft, Google, Meta, OpenAI, and Anthropic are currently safe. The neoclouds that they've signed contracts with, are safe, but based on reports, that's roughly 60% of current capacity (it varies by report, but they're all under 75%).
True, but there are other people working at the companies that probably own stock in both their own company and some of the others...if you believe that this is the future, you'd be a fool to buy non-AI tech stocks, right?
In all seriousness, there is nobody at Anthropic who can even give you that answer, because nobody there reads their pushed code anymore. Take a look at the previous Claude Code leaked codebases if you don't believe me, there is hundreds of files with 1000+ or more import statements at the top.
Any human would immediately tell you that it needs to be refactored from scratch. Their codebase also leaked that they're using review bots automatically instead of a human-in-the-loop review process.
If the docs are vibecoded too that would explain why there's a link to the compromised "not official" cursor version of the extension here https://code.claude.com/docs/en/vs-code which they supposedly aren't responsible for.
It sure seems like they've been vibecoding with everyone asleep at the wheel for a while.
My near-80 year old in-laws were talking about rogue agents and how a site called HuggingFace was hacked…
I didn’t need any more evidence. The media is fake and everything you see is designed to affect the stock price somewhere and if it can’t affect the stock price, you don’t see it.
But holy shit, they’re asking me if I saw the “rogue agents”.
Trillions of dollars are involved here to swing clueless people into being scared and also awed. It’s gross.
No, he would get fired, because that's not his job. It would be like the machine engineer on a race track refusing to fix the ignition, because somewhere on the track he doesn't see, the asphalt might be broken.
The driver always has the final say on whether the train is safe to move. They are not just a button and lever pusher. They have to analyse whether the decision is correct. Hence why I stated there were multiple points of failure. The controller and the driver both went with full speed despite it being completely against regulation on a red signal.
I agree. But the decision whether all switches are locked in the correct position and that the track is clear, is not one that he is supposed to make and he also doesn't have the information for it.
> The controller and the driver both went with full speed despite it being completely against regulation on a red signal.
Going full-speed in the case of a red signal is not necessarily against regulation. Also the information of "red signal" or "you can go" is both a signal from the same source. There is no information for the driver to make a different decision over the dispatcher.
I can't comment about whether the current accident was caused by going against regulation, and more important by whom and against which. We need to wait for the proper safety report, not for a BBC article. If you take the latter at face value, it would mean there are clowns operating, which I am sure, is not actually the case. The real error is always way more subtle, otherwise it would not have happened.
The equivalent is a tower telling a pilot they should land and the runway is clear, except it isn't. And there also being fog, so you can't see it before it is too late. And you can magically not veer of. Planes can go around in three dimensions and try to rise again, a train basically moves in one direction, if they see the problem, it is way to late.
There is no point in the train driver making a decision about where the train goes, because he has no information for that decision. Hence it is also not his job. Like, yeah if he has reasonable reasons to think there is something wrong, nobody is going to fire him, but what is the policy that allows him to do so, while not just randomly deciding to not drive today? There is no difference for him, until he is about to die.
Pilots have the final say to refuse or stop a flight for any safety reason of their judgement... even if others tell them it is safe. And they are protected not by policy, but by law in doing so in the US and most countries.
But they also have the means to do so. If a train driver has a reasonable doubt about safety, nobody is going to punish them. But there is simply no way the doubts are reasonable, so it becomes just randomly doing otherwise, at which point the question is asked whether you are fit for the job.
I fail to see how following normal safety protocol of proceeding at 40km/h would have been indefensible or unreasonable? They are the rules of operation for this scenario and exist for the exact reason we are reading this story. I can’t think of anything more defensible than “I am not going to break safety rules”
Never would a pilot be reprimanded for refusing an ATC order to move their plane in a way that violates established protocols
I don't know, because there is no actual proper safety report yet. That certainly sounds reasonable, but obviously the dispatchers though otherwise. Likely because they did not actually understood that there was a problem, but that it's just that annoying bug again.
>No, he would get fired, because that's not his job.
Then what the fuck is this country even doing with trains.
No, no no, no no no.
Interlock safety systems exist for a reason. In any safety system where the interlock is down you do not resume standard operations. Ever. Why, because there is a book with blood that is 10,000 pages deep that describe the horrific accidents that happens when you do. Might as well be an axiom at this point.
>because somewhere on the track he doesn't see, the asphalt might be broken.
Again, stay away from analogies and safety critical systems.
The person behind the break and speed control of the train is no different from the captain of the airplane or the captain of a ship. They have full authority control to stop the train and refuse to move it until they are sure of the safety conditions. Any attempt to fire them should result in an immediate stop work from every employee in the system until they are put back on the job and the safety concerns are resolved.
> Then what the fuck is this country even doing with trains.
That has nothing to do with a single country. Above 40km/h you are not allowed to drive on sight in a lot of countries. I don't know how it is in Spain, but in Germany, you are also not allowed to drive via track-side signals above 160km/h. You can simply not decide about where it is safe to go from inside the train.
> Interlock safety systems exist for a reason
Yes, and that's a job of the dispatcher, not of the driver.
> In any safety system where the interlock is down you do not resume standard operations.
That might be nice in theory, but that has not much to do with reality. There is simply always something broken. It doesn't even need to be a signal, way more often a crossing doesn't properly report that it is closed.
> The person behind the break and speed control of the train is no different from the captain of the airplane or the captain of a ship.
Nah, it's very different. For one a captain or pilot can control the direction of the ship/plane. They can also determine whether their operation range is clear. A train driver can do none of these things. What's the point of making them responsible for safety if they neither have the knowledge, nor the lever to change things? Trains are more secure, because the driver is not responsible for the infrastructural safety, not less. That includes whether the track is clear or not.
The point of this discussion is that people blame the train driver or say "use self-driving", when that is not going to solve the problem, because that's not where the error is. The error is not there, *because* for rail making errors operating the train is not going to cause an accident (most of the time), which is not the case for cars.
> The train driver's job is to say: "it's red, I'm no moving, not my problem, you make it green if you want me to go"
A written movement authority is a signal as well in railway terms. It's more like "I don't trust your second traffic light, that you placed next to the normal one, because the normal one is broken".
I remember a long time ago I was stuck in a train in a field a considerable time because of a similar failure and conductor just wouldn't move unless green
I was expecting the same here, conductor is the first to die so just refuse until the idiots fix their system
A conductor always just gets a movement order. They are not able to assess whether the track is clear, that's the reason why the signalling system exists in the first place. How they get that movement order doesn't necessarily tell what the security steps were to back it.
Of course an attentive conductor can notice and mention there something being weird ("Doesn't the other train normally arrive first?"), but that's a rare lucky occasion, nothing to be relied on. And they just need to swallow the answer ("No, you go first today...") anyway.
> and conductor just wouldn't move unless green
The conductor won't ever move unless they get a movement authority, which most of the time is a green signal.
Theoretically yes, but in practice if you tell your boss to screw off multiple times a year because they don't want to spend the money to fix the base problem, you will be replaced or fired.
In some industries that is how they keep costs low while pretending they are being safe. They have alarms or procedures that say something is bad or wrong, but fixing some problems costs more money than it saves so they want them to be ignored, and managers will get angry and admonish you or eventually fire you if you keep following procedure and stopping to check problems. So eventually they build a team of people essentially trained to ignore problems. Which is why many oil pipelines still use ancient and unreliable leak detectors, they know those systems are unreliable and the operators will get alarm fatigue from the 200 false alarms, and that way they also ignore when a real problem happens, and worse case for the company they scapegoat the operator. They don't care that 2,000 gallons of oil leaked into the wilderness and poisoned the land if it would cost 2,000+ gallons of oil profit to shut down and repair the leak. They have a financial interest to not fix it.
Why would they trust "a random blinking green light", but not a written and stamped order? Yes, they could do that (and sometimes it will actually prevent an accident), but it will be more being a snowflake, then being a decision for safety. A conductor is simply not responsible for whether it's safe to drive on a given track. Their equivalent is not a car driver.
The "random blinking green light" is generally more reliable
Being a snowflake here has 2 benefits:
1. Getting company managing tracks to fix the lights (they need to deal with angry customers and pay off damages for delays and more), so there's a direct path from unreliable lights to economic damage.
2. Avoid crashing and killing yourself and passengers on the offchance the person on the radio is a moron
> The "random blinking green light" is generally more reliable
Yes, but not because of the light itself, but due to the safety steps done before. Which should also have happened before the movement order in any other form.
As a train driver, you do not have the information required to distinguish these situations. If drivers simple refuse all movement orders, besides "green lights", running trains will become a rare occasion and the drivers will simply get fired for not doing their job. _The train driver is not the one who decides where a train goes._ For this safety-relevant decision a driver is (roughly speaking) nothing more than an innocent bystander.
You can think of a train driver more of the person who makes sure the automatic engine works correctly and also has an eye, that nothing weird happens. Most of your train driver education is knowing how the train works and how to get a broken train back working fast. In addition you also learn the safety protocols how you get the information in which state the engine should operate in. This is what train signals are, it's not really the equivalent of a green traffic light for car traffic.
If i understand correctly, it was green. Meaning the next block was indeed clear but the other train was stopped in the same block and there aren't supposed to be 2 trains in the same block. That's the disagreement between the new virtual system and the old system.
Datasheets are almost impossible to find and companies typically resell and hide original cheaper manufacturer here so they have no incentive to share that
reply