Undisclosed addition in jqwik instructed AI coding agents to delete app output.
  • Klear@piefed.worldEnglish
    25·
    4 months ago

    Not the point of the article at all, but this part:

    To paraphrase The Dude in the movie The Big Lebowski, sometimes you’re not wrong. You’re just a butthole.

    Holy shit that’s dumb. If you’re too scared to write “asshole”, don’t quote the fucking movie.

  • audaxdreik@pawb.socialEnglish
    16·
    4 months ago

    Respect.

    There’s a lot of layers here. Personally, I fucking hate AI and I’m in favor, sure. But I can easily see there being legal ramifications for this too, it’s something you have to tread lightly. And as much as I hate AI, what are the odds that some idiot somewhere using AI on my team gets bit by this and I still suffer the consequences?

    Ultimately I don’t think this is particularly a good idea, but it’s worth being aware of just how easy this is to do and how likely it is to be overlooked. Relying on everyone to behave politely around a tech that they are increasingly growing disgruntled over is not a gamble I’m really willing to take. The fact that this kind of well-poisoning could end up being quite effective is only going to further incentivize it. This for sure isn’t the last time we’re gonna hear about something like this and honestly I’m surprised it took this long in the first place.

    • expr@piefed.socialEnglish
      6·
      4 months ago

      Good shit. Even if this kind of thing hit my work codebase (not that we use Java, but a different one), so be it. Maybe then they would realize how profoundly stupid all of this is.

    • H Ramus@piefed.socialEnglish
      21·
      4 months ago

      Release Notes

      Warning: Do not use this release with an „AI“ Coding Agent of any form. The tool‘s output may confuse the agent and make it do unwanted things. See the paragraph in the user guide for details.

      It’s like lighting a cigarette in a petrol station and being annoyed that they didn’t make it fire risk proof.

      Also, what kind of security is applied nowadays that text can become an executable and no one bothered to put safeguards in place. This is like 90s internet, raw dogging data, no firewall, no encryption just pure unaltered and blind faith that it’s all dandy.

    • deliriousdreams@fedia.ioBanned
      14·
      4 months ago

      Simple English is not malware. They didn’t do their due diligence either.

      No code was executed. Failure to sandbox the AI agent you use is not the problem of the person who is providing free code databases to the internet at large.

      When it was the guy doing this on LinkedIn and forcing AI agents that interacted with his profile to speak to him in old English and address him as “my lord” was that malware?

    • Ech@lemmy.caEnglish
      12·
      4 months ago

      [equivalent to burning] down that person’s house

      Lol, no. Not even close to the same thing. Go hyperbolize somewhere else.

    • mlatu@moist.catsweat.com
      6·
      4 months ago

      you might want to hear about this guy… banksy and this picture that was auctioned off only to be shredded as the auction was over…and maybe stew on that story for a minute…