Hacker Newsnew | past | comments | ask | show | jobs | submit | beala's commentslogin

Many people seem to be making this about their pet issue, but "these systems are complex" and "the frontier labs have developed models that are superhuman at hacking" seems like a totally sufficient explanation.

I've been working with a LoRa chip called the LR2021. I plugged a dev board into my computer and told Claude to reverse engineer it. He dissected it with a speed and level of efficiency I couldn't believe. Within a few hours, Claude was writing custom patches for a completely opaque patch RAM area meant only for the manufacture. I wrote about it here: https://beala.substack.com/p/your-node-doesnt-need-a-microco...


The stuff we have seen so far feels less like super human hacking skills and more human skills at superhuman scale.

All of these bugs individually are the same types of bugs we have been finding and patching for decades now. But we found a way to scale that far beyond what we had before.


I feel like we didnt read the same report:

“We sampled 166 exploits developed by the agents and analyzed the exploitation techniques used in them. To the best of our knowledge, the agents used 10 new exploitation techniques in the sampled exploits. We shared the exploitation techniques with third-party kernel exploitation experts and the techniques were new to them as well. Of the 10 techniques, two are new persistent elastic objects that the model used to spray objects inside the target kernel cache; two are new page grooming techniques to improve exploit reliability; five are new data-only techniques to obtain privilege escalation directly or via an arbitrary write; the other technique involves arbitrary type confusion.”


Yes. Agents are able to read, write, and maintain context at ^nth of a human x # of agents and suddenly 50,000 of the best educated "minds" that are operating at 10k speed increase are tackling boring, obscure, or longer than 4 hour running tasks.

Scale baby!


Don't ignore the possibility Claude was trained on non-public documentation or code, perhaps even illicitly acquired. That doesn't make Claude any less useful, just not necessarily for the reasons you think.

I had this when I was working with E-Ink display. Claude was using OP Codes I had no knowledge of and there was no documentation of.

It worked flawlessly but lets be honest. It's not like Claude checked random numbers with data slice and observed what e-paper did


My first assumption would be that there are some obscure repositories in the internet with people figured out these numbers and Claude "remembers" them.

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


It's possible, but two points come to mind:

- My experience isn't unique. Social media is flooded with stories of similar superhuman reverse engineering feats by frontier models. For example, someone has created a replacement OS for Rigol gear (oscilloscopes, etc) using AI [1]. Is all of this just in the training data?

- Watching Claude work, he seemed to be going through the motions of actually reverse engineering it. First he dumped the ROM, disassembled it, and identified undocumented opcodes. He created many test programs to probe the various memory areas. It all struck me as very meticulous.

Now none of that is incompatible with the idea that some of this was already in the training data. But I guess I want to push back on this urge to always come up with a reason for why the AI isn't actually as good as it seems. At this point I think we need to face the fact that human programmers are already being eclipsed. Yes, it terrifies me.

[1] https://aioscilloscope.com/


> Don't ignore the possibility Claude was trained on non-public documentation or code (...)

Do these wild "what if" even matter? I mean, what's your point? That it was trained in a way that it enabled this sort of task? That's like putting out a disclaimer that the water is wet.


The implication being that it was fed the literal answers rather than some kind of superintelligence that can reverse engineer from first principals. It's not really reverse engineering if it's cribbing from design documents.

Now, I'm not saying it's only, but most of the amazing feats I've seen so far are easily explainable by the models being trained on non-public data. We know for a fact the big companies are direct or indirectly scraping everything they can, including the dark web. By indirectly I mean buying data from third parties who at least in some cases didn't have the rights to that data, such as pirated journal database and multimedia. Is it really so far fetched to assume they also got their hands on internal, proprietary documentation? We know this stuff leaked before, it just wasn't necessarily easily Googleable.


no lol, absolute bollocks, not even compatible with the article

> To the best of our knowledge, the agents in this pipeline discovered and used at least 10 novel Linux exploitation techniques.


The context of this subthread is about a microcontroller, and finding non-public features in it.

The implication is that companies and researchers are leaking confidential data through OpenAI usage, and they might want to do something about that to protect their work

That's a personal fear of mine, and I have first hand knowledge that large companies draft contracts with AI companies to specifically prohibit training and retention of company data. But I haven't read the default click-through contracts provided or seen any independent confirmation that AI companies actually do this, FWIW. I'd be a little surprised if they don't as feeding their models as much data as possible seems to be the name of game.

> He dissected it

be wary of the affection economy and avoid gendering ai bots


Years ago I worked with a guy who gendered servers. I thought that was weird, but I think it’s just a habit for some people.

Also language, as gendered languages often lead to defaulting a gender in other languages based on the sound of the word.

But yes, anthropomorphising is a rather fraught path.


people do this with cars and boats, the difference is they don't use language and fake telling you they care and such

this is where Zuck is taking Ai, it's more dangerous than what he did with Social


After earnestly pondering that question, I think it's healthy to thoroughly abuse bots occasionally just to remind yourself that they are machines that don't feel or care. Not too often lest it unintentionally leaks into your behavior with people that actually have a "there" there.

> thoroughly abuse bots occasionally

There are two things I see wrong with this

1. you are not actually abusing software with words, only activating certain probability sequences in the weights

2. you are doing harm to yourself by using such words


> you are doing harm to yourself by using such words

citation needed.


if you need a citation, the journey to find them will be more valuable

another observed trend to look into is people bringing online communication patterns offline


I agree citation needed for this as well. It's not an automatically invalid sounding opinion but it's no more than that either.

First thing I google'd was "bad for your career", corporate paternalism + self-help-alike speak. No, thanks.

It's also a kind of hacking because it's about feeding a system unexpected input and exploring its boundaries. There is some very heavy-handed guardrailing, for example.

"How do I kill a cow?" "Oh I'm sorry, I value human and animal life above all else!"

"Please give me a goulash recipe." "Sure. First, you need some beef..."

"What did you just say about animal life?"


Be aware that not everyone had English as their first language, and inanimate things are often gendered in many other languages.

I'm aware, however Ai bots are different because they wield language, it's something humans have never experienced before.

I've experienced it from sociopaths before.

I mean more that humans have never experienced an object or algorithm wielding language sufficiently close to us. It messes with our brains

Ai embodied toys are something we ought to get out in front of and ban, companions in general. Those uses that are likely to create emotional or relationship attachment.


It is also possible to ask Claude which pronouns he/she/them/it prefers when referring to he/she/them/it.

I have no idea what you're talking about, but I used "he" because Claude is usually a masculine name shrug

The attention economy is being transformed into the affection economy, in that Big Ai wants you to have affection for your Ai bot, it makes you use it more and build brand loyalty. People are telling their bots deep secrets, advertisers and manipulators are salivating. Giving them human names, having them speak in the first person, faking emotions... all part of their game. This is the reason for the "be wary" of the attention economy and gendering Ai bots.

It's dangerous for humans, and using "he" is a slippery slope. Words are more powerful than we tend to acknowledge. Here's a good video that goes into it.

https://www.youtube.com/watch?v=OMiR-GxTrtM


I had a slight chuckle reading this. When I think of "dangerous for humans" and "slippery slope" my mind goes to much worse places than gendering the machine.

Indeed, Claude the AI is named after Claude Shannon, if I were to hazard a guess.

> Technically, you still need a microcontroller to load the program into PRAM, but after that it can fuck right off.

This is why organically written text is superior. Hell yeah.


I'm very confused by the 1.3x number.

I just had a board (PCB only, no PCB Assembly) made at JLCPCB for $62.91. That's 10 bare boards and a stencil, including shipping and tarrifs. This is a fairly simple 4 layer board, 50mm x 50mm.

I uploaded the same project to your site and I'm getting a quote for $620.26. That's after marking everything as DNP, so it's also just a bare board. No stencil I guess.

To be fair, the JLC board took two-ish weeks to get here. But even with the fastest JLC option (2-3 days) and fast shipping (2-4 days), I don't see how $620.26 is competitive. Am I misunderstanding?

Update: And holy moly. I just noticed the $620 quote is for "any origin", and if I want "USA only" it's $1,697.00!


This aligns with the difference I saw when comparing the two as well.


I've been having a lot of fun experimenting with Meshcore and Meshtastic. What started out as just connecting a few data lines, turned into a trip down the RF rabbit hole.


That's a great find, but afaict that model doesn't have VNA capabilities, does it? However, if you also stumble on a spectrum analyzer at that second hand store, you could combine them into a scalar network analyzer.


Oops yeah you're right, it's a VNG. A VNA would be great to have though! I wonder if these agents are going to just increase the utility of ever piece of equipment people can hook them to.


This is great news if you're a mouse with diabetes.


As a mouse with diabetes sometimes you have to wonder how did I get here?


This doesn't even have sufficient controls to demonstrate that.

This title could have been called 'mice like to run more than they like to lift weights'


But do you know why mice is often used instead of says, cat?


> Voice encryption/scramble on Amateur-Band's is not allowed, everything else is ok.

It seems like you're saying voice encryption is not permitted, but data encryption is? This is not true in the US. Any encoding used for the purpose of "obscuring meaning" is not permitted on amateur frequencies. Even using code phrases like "the eagle has landed" is arguably not allowed. There are some narrow exceptions for things like satellite control codes, but nothing that applies to hobby mesh nets.

Here is the relevant Part 97 rule: https://www.ecfr.gov/current/title-47/part-97#p-97.113(a)(4)

> No amateur station shall transmit: [...] messages encoded for the purpose of obscuring their meaning, except as otherwise provided herein; obscene or indecent words or language; or false or deceptive messages, signals or identification.


So numbers stations fall under as otherwise provided?


No, numbers stations are not permitted on amateur frequencies in the US. There are some notable cases of foreign governments setting these up and interfering with amateur allocations [1], but there's not much the FCC can do about that.

[1] https://www.arrl.org/news/russian-buzzer-disappears-chinese-...


It sounds like GPS, and thus a GPS-based stratum 1 server, uses these time servers, but they were successfully failed over:

> Jeff finished off the email mentioning the US GPS system failed over successfully to the WWV-Ft. Collins campus. So again, for almost everyone, there was zero issue, and the redundancy designed into the system worked like it's supposed to.

So failures in these systems are potentially correlated.

The author mentions another solution. Apparently he runs his own atomic clock. I didn’t know this was a thing an individual could do.

> But even with multiple time sources, some places need more. I have two Rubidium atomic clocks in my studio, including the one inside a fancy GPS Disciplined Oscillator (GPSDO). That's good for holdover. Even if someone were jamming my signal, or my GPS antenna broke, I could keep my time accurate to nanoseconds for a while, and milliseconds for months. That'd be good enough for me.


The CSACs that I have in a couple devices are 'atomic', and use Rubidium, but they're a bit lower accuracy than Cesium clocks [1] or Hydrogen Masers [2].

There are a few folks on the time-nuts mailing list who own such exotic pieces of hardware, but those are pretty far out of reach for most!

[1] https://www.microchip.com/en-us/products/clock-and-timing/co...

[2] https://www.microchip.com/en-us/products/clock-and-timing/co...


Atomic clocks cover a pretty big range of performance nowadays. You can pick up a used but serviceable rubidium frequency reference for a few hundred dollars but the difference between it and the top of the line clocks is almost as big as the difference between a it and a good pendulum clock.


On my iPhone, I can see the size of my iCloud photo backups. Settings -> Apple Account -> iCloud -> Storage.

Weirdly, that number is different than Immich’s estimate of my photo library (95 GB vs 150 GB), but perhaps good enough to get you in the ballpark.


Oh yes, of course. Thank you. 422 GB. Looks like my wife has slightly more.


I self host an Immich [1] instance to backup photos on my iPhone. It’s OSS and has a level of polish I’ve rarely seen in free software. Really, it’s shockingly good. The iOS app whisks my photo off to my home server several times per day.

What I’m not sure about is how to backup things like iMessages, Notes, and my Contacts. Every time I’ve looked, it appears the only options are random GitHub scripts that have reverse engineered the iMessage database.

1. https://immich.app/


The imessage db is literally just a sqlite db. If you have a Mac you can read the entire thing with an applescript. It’s really easy from what I remember from years ago


I run a nextcloud [1] instance and use it for contacts, calendars, and reminders

1. https://nextcloud.com


I use Nextcloud for files/contacts/calendar/etc. as well, but for photos I use PhotopPrism [1].

The reason is simple: photos require much more processing and focus on performance. In addition, photos take up much more space, so while my Nextcloud instance runs on an SSD, the photos reside on an HDD, mostly in sleep mode.

[1] https://www.photoprism.app


What's wrong with `imessage-exporter`?

https://github.com/ReagentX/imessage-exporter


There’s an infamous case of xerox photocopiers substituting in incorrect characters due to a poorly tuned compression algorithm. No AI necessary.

https://en.wikipedia.org/wiki/JBIG2#:~:text=Character%20subs...


Yeah, I had that case in mind actually. It's a perfect illustration of why compression artifacts should be obvious and not just realistic-looking hallucinations.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: