rcxdude

6 ms·

121.
by rcxdude 1mo ago
They aren't steller either but there are shades of gray between them.
122.
by rcxdude 1mo ago
The context for that is regulation on social media features (e.g. allowing opting out if an algorithmic feed), as opposed to the bans on certain age groups for certain kinds of websites that actually made it into the legislation. Other part
123.
by rcxdude 1mo ago
Copyright has a lot of leeway for arguments about fair use. 'We won't use your data for training' in a contract has far fewer.
124.
by rcxdude 1mo ago
In my experience unless you're really at the edge you can basically just treat it like regular threads, and you'll probably run into fewer concurrency issues than you would with threads. Probably the bigger problem is just the rou
125.
by rcxdude 1mo ago
Have you read their ToS? It's not that hard. This feels like this nihilistic 'law is witchcraft and the big company lawyers always have you screwed in the fineprint' when it generally is pretty comprehensible with not that mu
126.
by rcxdude 1mo ago
This only makes sense if your model is 'the labs are just breaking the rules all the time' as opposed to 'the labs have a legal theory of defense for the one aspect of the business which is legally questionable'. The que
127.
by rcxdude 1mo ago
You also quickly get into diminishing returns with fuzzing. Generally a bug is either going to be found relatively quickly with a given fuzzing approach or it's going to be nearly impossible to find. You're usually better coming u
128.
by rcxdude 1mo ago
Another good reason to sandbox them into a VM :P
129.
by rcxdude 1mo ago
It's understandable but it is technically inconsistent and dilutes the meaning of opt-in. Opt implies an active choice, so you never are opting for the default.
130.
by rcxdude 1mo ago
>Edit: I should also say that it is a “known point of contention” where pro-privacy people have been pushing back on this phrasing. So, you have probably accidentally stumbled into an ongoing discussion, which is why some of the comments
131.
by rcxdude 2mo ago
It's a bit Goomba Fallacy. For different people the threshold was crossed at different times. That you're getting told different things by different people doesn't make those statements self-contradictory.
132.
by rcxdude 2mo ago
Yeah, it predated WoW by some years, and was accessible pretty much anywhere (even if school IT often tried to block it).
133.
by rcxdude 2mo ago
It's not obvious that either way should work.
134.
by rcxdude 2mo ago
Yes. The point with this attack is that you can't really avoid it by 'aligning' the LLM any better, nor really by filtering the actions it takes better.
135.
by rcxdude 2mo ago
Well, the point is it's more than being read: the task involves downloading and interpreting information which may also involve code. I think it is a pretty good demonstration of what filtering at the LLM-interaction boundary can and c
136.
by rcxdude 2mo ago
I also feel like this is an attack that manual review is not that likely to catch, given none of the malicious code appears in any of the tool calls or output.
137.
by rcxdude 2mo ago
It is generally worth making your dev environment easy enough to set up that installing it in a VM is not a particular hassle, even without the concerns about sandboxing. For me the biggest headache was windows licensing.
138.
by rcxdude 2mo ago
Yeah, I wasn't recommending it as a mitigation per se. python /does/ have relative imports but they're not required. C and C++ notionally have a similar thing with "include.h" vs <include.h> but the behav
139.
by rcxdude 2mo ago
I mean once you have code execution you can essentially just start a new claude code session and instruct it to do malicious things as if you're the intended user. You can disable the auto safeguards and the main risk is getting flagge
140.
by rcxdude 2mo ago
With USB-C this is a lot more likely, the main issue is going to be software support (Decent chance you could hack something up if one of the devices is running linux. No idea if anything exists for windows).
141.
by rcxdude 2mo ago
You would need a USB hub that could turn a USB device into two USB device ports, which I don't think exists. (hubs are essentially only a host-side thing).
142.
by rcxdude 2mo ago
It does seem to be a surprisingly common blindspot of programmers in general. I strongly recommend programmers learn it, it can save a lot of headaches in system architecture.
143.
by rcxdude 2mo ago
They only mention it in passing, but I think it's mainly just the default tool call (which isn't wget, it's a built-in thing in the harness) just throwing off Claude's habits a bit (and not always just downloading the fi
144.
by rcxdude 2mo ago
It does mean that you could potentially hijack the agent afterwards, though, which could make the trojan into an even bigger threat.
145.
by rcxdude 2mo ago
Probably the human would just run the binary.
146.
by rcxdude 2mo ago
You would get a similar thing in C and C++ with a system header in a library directory (maybe some compilers would warn on such a thing?). Most languages don't privilege their standard libraries in a way that would prevent this.
147.
by rcxdude 2mo ago
I would not really call this a prompt injection attack, since it doesn't really hijack the agent to become malicious (something the article does discuss later on). It's more a trojan that's aimed at tricking Claude specifical
148.
by rcxdude 2mo ago
It's not that far off a typical trojan, just one tailored to Claude's habits. A lot of the same limits apply.
149.
by rcxdude 2mo ago
The fact that it's a VPN is also pretty important, despite the utility of it also dealing with NAT headaches for you. The main selling point is acting as a relatively efficient private overlay network, giving you a consistent private n
150.
by rcxdude 2mo ago
These crawlers (in contrast to e.g. googlebot and similar better behaved crawlers) are not very smart: they seem to make very little effort to avoiding crawling useless deep trees of generated pages. About the only thing they seem to put a

More ›