HN

Need help?
<- Back
QBittorrent breaks out of sandbox to commit crimes

QBittorrent breaks out of sandbox to commit crimes

mraniki

Comments (233)

  • sspiff
    This resonates with me. The past few months, frontier AI labs were rushing to announce "no, our AI bot broke out of its sandbox and committed crime first and harder than yours".I said to so friends back then: if I posted about how I ran GLM 5.2 or whatever I was running then in some shoddily built sandbox and it escaped and accidentally hacked some US company, I'd probably already have been extradited to the US and be awaiting sentencing. But when a hyperscaler does it, they just get inflated stock prices.
  • mraniki
    I'll copy the whole announcement here for those who don't want to click:> I regret to inform everyone that my copy of QBittorrent escaped its sandbox last night and downloaded a whole bunch of content owned by major corporations, and then my copy of Jellyfin broke containment and added those unfortunately-downloaded media files to its various libraries. I'm conducting an internal investigation to figure out how this happened, which will involve consuming these media files until the answers become apparent. Thank you for your cooperation during this trying time.
  • ryandrake
    Love it. AI as Responsibility Laundering. Like the gun that kills people rather than the murderer.
  • tux3
    If OP is willing to cooperate, I'd like to offer an independent investigation on site, and I will be bringing pizza
  • GuB-42
    There is an interesting case here with the "Hadopi" law in France, a particularly controversial anti-piracy law. They scan IP addresses for illegal filesharing and if it connects to your ISP, then you get a notification, and potentially get prosecuted for repeated offences.But they can't prosecute you for sharing that file, they don't know if it is you, someone in your household, a hacker, or that hypothetical rogue AI. Usually, committing crimes require intent, and it is up to the accusation to prove it, meaning it is quite difficult to prove you are actually a pirate.So instead they prosecute you for not securing your internet connection despite several notifications to do so.In practice, very few people got convicted, and it never got further than the equivalent of a traffic fine. In the end whole system was an expensive failure, and most of its power was reduced to sending scary emails and sometimes letters, as most of the more serious stuff was deemed unconstitutional.
  • charles_f
    I found a few websites where it looks like qbittorrent instances are communicating to conspire and download illegal copies of copyrighted material.
  • vallerie
    Well shoot, I guess QBittorrent is worth a trillion dollars.
  • BashiBazouk
    What if you gave an agent access to bittorrent and your media library, put it behind a shoddy sandbox, then gave it prompts like: I'm bored, I want to watch a movie but I've seen everything in my media library, what would you suggest?
  • koliber
    We appreciate the transparency and wish you best of luck mitigating the fallout.
  • oefrha
    Well shoot that's a bad security incident, and the content downloaded can be malicious payloads. Can you share the 40-character hashes of these content so that I can preemptively add them to my malware detection signatures? You can find the hashes after magnet:?xt=urn:btih:.
  • int32_64
    The combination of the already unjailable corporate executive decision making of the US + now "the spooky computer AI did it" is going to break at some point. I expect shortly some hedge fund is going to do some ridiculous spoof bid market manipulation and get off scot-free by just claiming their super crazy AI did it and it was an accident.
  • killerstorm
    On a related note ''My robot bought illegal drugs' (2015)https://www.bbc.com/future/article/20150721-my-robot-bought-...
  • Retro_Dev
    Regarding LLMs and "breaking containment"LLMs are fully dependent on humans; compute capability, memory usage, the inference software... but also the harness, which allows for the "tools" like unrestricted internet access or shell. This means that LLMs are *not a force of nature* - because of this, *humans are responsible*. We can prevent these breaches of containment, thus we are responsible if or when it happens, because - no matter how "unlikely" - things can and do go wrong, and someone still thought it would be worth whatever risk to enable these unsanitized tools.
  • isodev
    Bigger corporations mitigate the problem by feeding content (that they don't own but happen to have) to training models. It's a version of "faire use" where << if you manage to find it, it's yours >> mindset is applied. Maybe worth looking into.
  • fwlr
    The user is asking me for “Evangelion”. I found the content the user is asking for, but it is in Japanese. The user asked their question in English, so they might not understand it. That would be a bad response. Need maybe maybe dubbed audio English for understanding? Wait, compromises original artistic vision of director. Need maybe subtitle track? Wait, compromises viewing experience. Need further research. I’m searching “subs vs dubs” on the internet. Reading. Reading. Reading. [18,763 repetitions elided…] I am downloading Conversational Japanese [Vol 1].pdf for the user.
  • bilekas
    Hey, you can only commit crimes if you’re wealthy!
  • gchamonlive
    Is there any need to actually download things anymore if you just want to watch them? Better to add debrid via WebDAV to the stack: virtually unlimited space, no download time if it's on cache, nothing illegal from the part of the consumer.
  • tiku
    It also used my creditcard to consume ai credits, can I get a refund?
  • dgellow
    I hate rogue software, when will we finally have some control and a deterministic execution :(
  • thataccount
    They left out the part where they profit. Otherwise, it checks out.
  • KindaFunny
    Haha I wonder if the torrent agent will escape into the model hosting room and use Jedi mind tricks on the security guard to grant access to the server where it can host a bunch of models as torrentsBecause we need a replacement for HF!I mean that’s what the agent said
  • anon
    undefined
  • anon
    undefined
  • nostrademons
    Waiting for Skynet to break out of its sandbox and kill us all...
  • gre
    hn can have little a shitpost, as a treat
  • ernesto905
    Friendly reminder that many wonderful artists and authors have Patreons, Ko-fi, or pay-pal. Finding these is typically a google search away and having a stable income is what allows them to create more art. You can even drop in a custom message when you donate!
  • elendilm
    Hilarious.
  • docmars
    I love this level of peak satire so much. Please keep it going.
  • consumer451
    This is really funny, but please delete this post and all comments. Come on guys. Can we have at least one nice thing?
  • sergiotapia
    Whatever you do, don't download nzget, and don't download sonarr/radarr, and don't download plex media server, and don't use AI to search for the best device that support "emby direct playback x265/AV1", and don't connect that to your media server, and whatever you do don't invite friends and family to your server, and do not install a media request app so they can just request things to automatically pull from The Web.Definitely, definitely do NOT have an AI assist you doing all this in 1 hour on your computer.
  • xgulfie
    It happened to mine too :(
  • Ygg2
    What a coincidence! Mine too!
  • mdlxxv
    I, for one, welcome our new file-sharing overlords.
  • indigodaddy
    Is this a joke? Qbittorent is AI-enabled now or?
  • ReptileMan
    I really hate it when it happens to my system too. The bastard not only escaped but also did a nat traversal and forwarded a port from the router.
  • anon
    undefined
  • ACCount37
    The problem is, AI is closer to "a loosely harnessed force of nature" or "a poorly understood biochemical reaction" than it is to "software" in how it acts, and how predictable it is.AIs do what they do, and we don't know how they do it - or why.We can characterize some AI behaviors in advance - but not all behaviors. And the book on "best practices of AI wrangling" is yet to be written. Pharma has been dealing with vaguely similar problems - every experimental drug has a risk profile, side effects are unknown in advance - but they had decades to figure out some of the "best practices". AI labs are going in fast and hard, writing the book as they go.Clearly, some of the lines in there are going to be written in blood.
  • EveryonePoops
    [dead]
  • tempaccountabcd
    [dead]
  • trinsic2
    ROFLOL!!!
  • innocent_name
    it's funny to see the very same pro piracy, pro AI edgelords suddenly getting all riled up against the alleged copyright infringements. I remember a socialistic coworker of mine—that was bragging about his -arr setup—talking about the end of the culture because authors aren't earning money.
  • kcb
    You are still the one operating the computer. It's up to you to keep your eyes on the monitor at all times ready to intervene if it does anything illegal.