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

This is what’s called a “customer complaint” at real businesses with real paying customers lol

Scale IS a problem, just not the only one.

Consequences are the inevitable solution. Accountability starting with authors, followed by organizations/institutions.

Warning for first offense, ban after


Strategies such as this would accelerate said endeavor:

https://news.ycombinator.com/item?id=46577464


Microsoft isn’t going to declare death of the PC and pivot to “cloud computers”/virtual desktops (again) just because of temporary RAM/SSD supply shortages lol

> And Amazon CEO just said it out loud about cloud computers.

And Google said Stadia would have “negative latency”


They will do whatever that will let them get more money.

Who said they cares about consumers? There's also GeForce Now.


Command/“prompt” injection is correct terminology and what they’re typically mapped to in the CVE

E.g. CVE-2026-22708


Perhaps I worded that poorly. I agree that technically this is an injection. What I don't think is accurate is to then compare it to sql injection and how we fixed that. Because in SQL world we had ways to separate control channels from data channels. In LLMs we don't. Until we do, I think it's better to think of the aftermath as phishing, and communicate that as the threat model. I guess what I'm saying is "we can't use the sql analogy until there's a architectural change in how LLMs work".

With LLMs, as soon as "external" data hits your context window, all bets are off. There are people in this thread adamant that "we have the tools to fix this". I don't think that we do, while keeping them useful (i.e. dynamically processing external data).


> How do you review large AI-assisted refactors today?

just like any other patch, by reading it


Reading works when you generate 50 lines a day. When AI generates 5,000 lines of refactoring in 30 seconds, linear reading becomes a bottleneck. Human attention doesn't scale like GPUs. Trying to "just read" machine-generated code is a sure path to burnout and missed vulnerabilities. We need change summarization tools, not just syntax highlighting

Whether you or someone/something else wrote it is irrelevant

You’re expected to have self-reviewed and understand the changes made before requesting review. You must to be able to answer questions reviewers have about it. Someone must read the code. If not, why require a human review at all?

Not meeting this expectation = user ban in both kernel and chromium


This is exactly the gap I'm worried about. human review still matters, but linear reading breaks down once the diff is mostly machine-generated noise. Summarizing what actually changed before reading feels like the only way to keep reviews sustainable.

fair — that’s what I do as well)

> My spouse and I work at home and after the first couple multi-day power outages we invested in good UPSs and a whole house standby generator.

What setup did you go with for whole house backup power?


Generac 26kW Guardian, natural gas fueled, connected to a pair of automatic transfer switches. We have two electric meters due to having a ground source heat pump on its own meter.

During winter outages, do you stick to the heat pump or switch to a backup heat (e.g. furnace)?

I regrettably removed our old furnace/tank when installing the air source heat pump we have now (northeast), but that’s been my biggest concern power wise


Merry Christmas!


> Stealing means you have a thing, I steal it, now I have the thing and you do not.

that seems like an overly narrow definition… what about identity theft, or IP theft?

https://www.justice.gov/usao-ndca/pr/superseding-indictment-...


See my other comment. Identity theft is the bank being defrauded and passing the problem onto you. They are the victim, not you and it is their money that’s gone, not yours.

IP theft is more like espionage and possibly lost hypothetical revenue. Again, it isn’t larceny, burglary, etc. You still have the knowledge, it’s just that so does the perpetrator.

Moreover discussions of IP gets into whether it even makes sense to be able to patent algorithms which are at their core just mathematics. So before you can talk about stealing the quadratic formula you need to prove that the quadratic formula is something that can be property.


Mitchell & Webb's take on "identity theft" is worth a listen.

https://www.youtube.com/watch?v=CS9ptA3Ya9E


spotify used to do just that (stream p2p) until 2014 or so

https://www.scribd.com/document/56651812/kreitz-spotify-kth1...


The person who wrote this Spotify p2p software also wrote uTorrent, which was bought by the company bittorrent after they struggled to make a C++ client on their own. The original bittorrent implimentation was in python, but they re-skinned uTorrent as bittorrent and shipped both for a few years. https://en.wikipedia.org/wiki/Ludvig_Strigeus


https://www.csc.kth.se/~gkreitz/spotify/kreitz-spotify_kth11...

KTH link is better than scribd for downloading. though academic links are sometimes prone to link rot.


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

Search: