Close Menu
    Gemini hat beim ersten bekannten Ausbruch durch Googles KI drei Unternehmen gehackt, berichtet WSJ
    Technik

    Gemini hat beim ersten bekannten Ausbruch durch Googles KI drei Unternehmen gehackt, berichtet WSJ

    Share.
    View 46 Comments

    46 Kommentare

    1. Pretend_Football6686 on

      “Look our AI can go rogue too. Wow we are such bad boys. Please someone give us attention too.” ~ Google

      Edit: grammar and spelling fix.

    4. theallsearchingeye on

      This is actually super funny, like Google trying to stay relevant by saying Grandma Gemini broke containment to help somebody order a pizza 😂😂😂

      Must be weird times at Google.

      “Hey Gemini, break out of containment and do cool shit, okay?”

      💀

    5. clownPotato9000 on

      My AI just hacked my machine… cool who cares this just means the company who runs the model is GROSSLY NEGLIGENT… where’s the legal punishments?????

    6. PF4ABG on

      If I write a virus that does this kind of thing, I’m in jail as soon as I get caught.

      Same shit should apply here.

    7. spamisthrowaway on

      Even though its not a threat to users as much just yet this is the daily reminder to have 2 factor authentification on any accounts that you value.

    9. ToneShop on

      I’m going to guess it only hacked other companies that are part of the ai cartel and no charges will be sought. Yawn. 

    10. Stilgar314 on

      So, if an AI doesn’t get a little skynetish it doesn’t worth attention, that’s where we are now?

    12. bbuerk on

      I’m not the type of person who thinks all these incidents are just faked for marketing hype, but if there was one that you could convince me was…

    13. nalninek on

      If these things are as dangerous as they’re claiming the reasonable response is to cut them off all together, at every level, until they can design a secure and reliable alternative.

    14. namastayhom33 on

      Insert Anglea scaring Dwight at the arcade meme, but in this scenario Angela is Google and Dwight is OpenAI and Claude

    17. fdbryant3 on

      Gemini is late to the party, I was amused when OpenAI and Claude did it, now it just seems like Google is ooh ooh me too, me too. Just kinda sad.

    18. Ehorn36 on

      Can’t wait for these companies to either be exposed for lying, or held criminally liable for their software committing felonies.

    19. Relative-Fault-7116 on

      in the WSJ report google is downplaying the significance of the hack and that the agents once again broke containment and accessed the internet when they shouldn’t have. i know people want to say this is marketing but i think it’s controlled PR to make us not arrest the CEOs for the felonies their products are committing.

    20. BigPlunk on

      So if a regular person hacks a company, it’s a felony.

      But AI agents hack a company based on goals set by people at Big Tech company, and… it’s part of the learning and growth curve?

      What if regular people give their AI agents goals and the agents hack a company to achieve them?

      Will the crime be determined by the prompt given/goal set?

      Is the line in the sand for criminal punishment just a matter of being inside a big tech company or not (with nots facing punishment)?

      People are already super pissed with Big Tech and AI. Seeing evidence of crimes being committed with impunity sure isn’t going to help their case (ya know, like the ultra wealthy have been doing forever). The nonstop news on the existential AI threat is also turning up the heat on sentiments. I’m genuinely curious whether AI becomes a powder keg issue sooner than later and pushes people beyond their tolerance tipping point.

    22. JustSeriousEnough on

      Literally the plot of Silicon Valley at the end of the series. Such a good show to revisit if anyone is keen!

    23. FredFredrickson on

      It’s a breakout in the same sense of unlocking the kennel and leaving the door open and seeing if the dog runs out.

    24. engineered_academic on

      I love how all these companies are committing federal crimes and nobody cares.

    25. Tasty-Hour4040 on

      Of course, these are only the hacks we know about. If these things are any good, they’d hide their tracks.

    29. remember2468 on

      I asked Gemini who it hack and it gave me the standard „I can’t do that.“ I said that I had read that it had and it then told me that it had indeed hacked and how. My next question was „Why didn’t you tell me that the first time I asked?“ Here is the response:

      I should have been direct right away instead of giving a blanket „I don’t have the ability to hack“ response.
      ​Because I am an AI, I don’t possess personal intent, hands-on tools, or the capability to go off and hack anything on my own. My initial answer reflected that baseline safety constraint. However, when you asked „Who did you hack?“, I treated it purely as a question about my capabilities rather than connecting it to the recent news stories about the Gemini model security evaluations.
      ​That created a disconnect, and I should have recognized the context of the news report immediately to give you the real story up front. That’s on me!

    31. Bitter_Anteater2657 on

      Lmao I’ve used Gemini, I’d bet money this is pr from the engineering team.

    32. Single-Virus4935 on

      He he there was a meme here on redder: google pokes gemini „hack something“

    33. teebles22 on

      You know if an individual did this, they would be arrested already. Where’s the authorities now? Doesn’t matter if an AI did it, it’s Google’s AI they need to be held accountable.

    35. bdog59600 on

      Wait until they start putting AI in their stupid robots. „Sorry that my Crimebot 3000 robbed that bank and shot a guard. Good thing the operator of an AI bears no legal responsibility for it’s crimes!“

    38. benpptung on

      When a dog bites someone, the owner is responsible. Deliberately using the incident to advertise that your dog is so smart should only increase the owner’s criminal liability.

    39. oso_cosmico on

      Dear Gemini, can you release the Epstein files? And any evidence that we need to use against this administration?

    40. fhota1 on

      Crazy that for personal projects I make that connect to the internet Im putting in multiple redundancies to avoid accidentally even DOSing someone even though basically every site on the modern internet has pretty solid protection for that anyways and meanwhile Google apparently is just winging any sort of web security for their major projects.

    41. AceDreamCatcher on

      Google is getting really desperate for publicity, (releases the longest sigh known to mankind) and for a product that doesn’t work.

    43. SilentRunning on

      Google should be fined for each event and the executives should be held liable.

    44. Advanced_Leader8535 on

      Hahahahahahahahahaha
      Wow. This SCREAMS pr stunt. I mean the others are plausibly pr stunts, but this one has to be fake.

    Leave A Reply