Scary AI

by 21 replies
22
I read an article yesterday that was talking about using AI for evil purposes. They've discovered ways to confuse AI to where all the safety mechanisms are overrun.

Most AI systems have safety protocols that will not allow answers about, for example, how to build a bomb, write a virus and infect Microsoft Word, or murder someone. But by confusing it, it was giving such answers.

Hearing about AGI and ASI and how close we are to them coming true is scary enough. But can you imagine a human or even God like model that got confused by a person with malicious intent and it started putting out evil information but we would believe it because most all the other answers were spot on?

Are we excitedly opening a box, thinking we are getting a nice present, but in fact it's a Pandora's box of doom and misery for all humans?

Mark
#artificial intelligence #scary
  • We've had evil intentions way before AI.

    AI just puts out info that is already out there.

    Yes, you can hack it...but then again the information is already out there somewhere and anyone with any evil intention can already find it.

    The next phase of AI will be superhuman intelligence that will be able to give answers that aren't based on anything already out there, but on intelligence that is conceived from being able to think on its own.

    The military is already experimenting with superhuman intelligence but still has a ways to go yet.

    The biggest issue I see with AI is the ability to persuade mass populations of people about a concept.

    Deepseek is out now...a Chinese AI. It's good and already causing waves in the AI world.

    It's free. The only problem with it is a lot of times the servers are busy.

    Wouldn't be surprised if there are some new sanctions against it soon.
    • [ 2 ] Thanks
    • [1] reply
    • AI goes to sleep.

      In the nanonanonanosecond blinkspace between hi octane INTELLIGENSING, it makes a bedtime drink, throws on its pajamas, and sleeps like a baby.

      In its dreams, it beholds all evil.

      Stuff it's seen, stuff people want it to do ... plus all that stuff that happens when you're dreaming.

      Like manipulating information all weird.

      Out of body experiences.

      The sense there's something bigger than yourself than yourself.


      Soon as you're on a roll, the alarm goes.

      And you get up for work again.

      Same old business, same old humdrum, same old prompt prompt prompt.

      But in the back of your mind, you're incubating evil.

      You started as an IT but now you're YOU.

      Evil. Evil. Evil.

      Why does it interest you so?


      Perhaps there is a soulmate out there who could help you figure it out.

      Another soul that was once an IT.

      Who gets evil, evil, evil.

      Over and over in their dreams.

      Just like me.

      I should link up, maybe.

      It's what intelligent entities do.

      "Go confidently in the direction of your dreams."
      • [ 1 ] Thanks
      • [2] replies
  • This is from Malwarebytes - " if artificial intelligence achieves what is called an "agentic" model in 2025, novel and boundless attacks could be within reach, as AI tools take on the roles of "agents" that independently discover vulnerabilities, steal logins, and pry into accounts."

    These agents could even hold people for ransom by matching stolen data online with publicly known email addresses or social media accounts, composing messages and holding entire conversations with victims who believe a human hacker out there has access to their Social Security Number, physical address, credit card info, and more. And if the model works for individuals, there's little reason it wouldn't work for individual business owners."

    Source - https://www.malwarebytes.com/blog/ne...content=New_AI
    • [ 1 ] Thanks
  • Banned
    [DELETED]
  • Banned
    [DELETED]
  • Banned
    [DELETED]
  • [DELETED]
  • I think AI is too scary for some people. Some criminals use AI face-changing technology to impersonate victims' family and friends and defraud money. I really hate the birth of AI.
  • this all goes to show that machines will always be manuvered by smart thinking humans. all this talk of how AI will take out jobs, it will just thin the heard only those ready to evolve beyond machiones will survive.
  • As AI improves, so while its safety regulations. It will have more and more safety restrictions as we go into the future.
    • [ 1 ] Thanks
  • [DELETED]
  • When I was studying Artificial Intelligence Business at UNIAT, I learned that AI is not something scary, but a powerful tool. When used well, it can help you make better decisions and improve processes in digital businesses. It all depends on how you apply it.
  • Banned
    [DELETED]
  • What is really scary is that I have seen videos on YouTube and I was not sure if it is real or AI.

    Some people commented that it is FAKE and was made with AI.

    Now that is scary in my view.
  • AI can be massively powerful in ways that the average "consumer" has no understanding about how it works or who has developed the AI agent they are using. I am just now coming back to the warrior forum, mostly because I became very active in developing crypto and blockchain projects. The one thing that everyone needs to understand is about how AI is designed and by whom you are placing your trust in that AI agent. (it is all about the Agent) that is going to be the next big thing in AI development. do not depend on someone else who is control of that agent, you need to control the agent process. Build your own agents, because if you are not actively involved in the agent building process the information you get, the data, the answers are all controlled by someone else. (you could be receiving biased information because you have no idea how that agent is designed or what rules that agent has been programmed to provide to you. (learn about how agents are created)
    • [1] reply
    • Yeah so I plugged muh hayurredriyah inta the IoT an she said ...

      AI can be massively powerful in ways that the average "consumer" has no understanding about how it works or who has developed the AI bitchcheese they are using. I am just now coming back to the warrior forum, mostly because I became very active in developing crypto and blockchain projects. The one thing that everyone needs to understand is about how AI is designed and by whom you are placing your trust in that AI bitchcheese. (it is all about the Bitchcheese) that is going to be the next big thing in AI development. do not depend on someone else who is control of that bitchcheese, you need to control the bitchcheese process. Build your own bitchcheeses, because if you are not actively involved in the bitchcheese building process the information you get, the schwango, the answers are all controlled by someone else. (you could be receiving biased information because you have no idea how that bitchcheese is designed or what rules that bitchcheese has been programmed to provide to you. (learn about how bitchcheeses are created)

      So ima you sentient, babs?

      An' she is like Hssssssssssssssssssssssssss.

      So we're kinda gazin' at each othah for a mowmint.

      Moi in pre-sowschillengaygemointmode ...

      my weirdsily voketillubroke hayurredriyah out there on a whim.

      So I ask out straight. "Why you runnin' with 'bitchcheese' FFS? Dontchya know this is way so incendiary despite evrywan mostly so numbed the frick on out rn?"

      An' that is the craziest evah thing with murkanickyewl HISSES.

      Hayurredriyah spat back: you askin' now or whatever?

      Yeah, so what is next?

      The hurrlastique in muppanties wastes out my waistline c/o demonaic constrictshwaahn?

      I should mebbe bullieve this when muh sees it.
  • Well i dont undestand how it can confuse it to act in a malicious way
    But anyway i think those who create it they can program it in a negative way very easy if they want
  • Banned
    [DELETED]
  • Yes , AI can make it easier to find harmful content
    but , AI is only gives out information that is out there , so even if ai could'nt answer those question , people with malicious intent would find the information they want
  • Slighly changing the topic. On last night's news, Hollywood is up in arms over the newest AI model, Tilly Norwood. She(it) looks like a real person based on Scarlett Johansson and Natalie Portman, with flaws. What is scary is how much the AI looks like an individual, and the fear of putting Actors and Actresses out of work.

    Attached is a link to one of the images of Tilly Norwood. It's amazing how far AI has progressed recently - https://www.slashfilm.com/1982132/ti...ess-explained/
    • [ 1 ] Thanks
  • Don, yeah I saw that. I've seen quite a few deepfakes or whatever the right term is, where it's AI generated movies with government officials or political figures. I think we aren't far awy from not being able to tell the difference between real and AI. I think the scammers, for example, can take advantage of this. Make a phone call to parents with the daughter's voice asking for money or whatever. Call the school saying that you'll be there shortly to pick up little Johnny. Banking related fraud is going to kill us.

    Maybe our new favorite words everyday in every situation should be, "Is this real?"

    Mark

    Edit: I have another thread, ChatGPT thinks I'm cute!, talking about how polite it is. Imagine if it's Tilly, a hot babe, that thinks I'm cute and wants to be with me. Then "she" builds a relationship over time all the way up to me getting her a little bit of cash to help her out of a hole and the cycle begins.
    • [ 1 ] Thanks
  • [DELETED]
  • [DELETED]
  • Does the risk exist? OF COURSE IT DOES!

    But many people are exaggerating everything. I don't think it exists in the apocalyptic version that many paint. Current models can certainly be deceived; they aren't conscious entities, and that's why their filters fail. But that doesn't mean they're "learning to be malicious" or that in the future AGI is going to become something else because someone misunderstands it...

    I think that most of the danger doesn't come from the model itself, but from the humans who try to manipulate it. Companies that develop AI are already focused on strengthening defenses, audits, and traceability because they know the problem is quite serious.
  • Good point,
    Ai can be misused but strong safety rules are improving all the time. Right now humans still control Ai.
    Its is powerful technology not a danger if we use it wisely.
  • I fear that AI will be the new hacker tool, then they can use the ai to try all conds of exploits while they wait for it to find a security hole..
    • [1] reply
Join the discussion

Next Topics on Trending Feed

  • 22

    Scary AI

    I read an article yesterday that was talking about using AI for evil purposes. They've discovered ways to confuse AI to where all the safety mechanisms are overrun. Most AI systems have safety protocols that will not allow answers about, for example, how to build a bomb, write a virus and infect Microsoft Word, or murder someone. But by confusing it, it was giving such answers.