The Register Home Page

back to article If you're not using AI to attack your own systems, your adversaries will

AI agents excel at hacking organizations, as they’ve demonstrated in real-life attacks multiple times over the past few weeks. They also expose a whole new attack surface for organizations trying to protect against both human and AI intrusions. As if defenders needed more worries to keep them up at night, agents introduce new …

  1. elsergiovolador Silver badge

    Attack

    So if you tell the Western models to attack, that might lead to your account being suspended.

    Great advice.

    Meanwhile foreign actors can use "uncensored" models often on par with the so called Mythos, or shall I say Meth-os.

    The whole idea of kneecapping of these models idea is dumb.

  2. amanfromMars 1 Silver badge

    Don’t heed clear leading messages? Then expect to forever play second fiddle to those who do.

    Y’all can’t say you weren’t told and warned your global executive systems and elite administrative officers are easy prey to new forces and sources and therefore it is vital that they up their game ...... if they can ...... or face particularly peculiar and extremely troublesome and irreversible consequences way beyond any effective means of future human command and control, power or energy .........

    amanfromMars 1 Mon 22 Dec 2025 08:13 [2512220813] ….. airs more on https://forums.theregister.com/forum/1/2025/12/19/doe_genesis_agreements/

    Re: Merlin the AIMagician and Mega Beta MetaDataBase Physician Dons the Full Santa Mantle

    The Gazillion Dollar Question left hanging heavily there, and here also, for all current running incumbent and/or incompetent leaderships of both the enthusiastically positively optimistic and the depressingly negatively pessimistic in humanity and societies down on Earth to ponder on and wonder at regarding the novel leading future courses which are daily and 0daily to be commandeered and presented via Mass Multi Media Operations servering IT and AI Solutions with Engaging and Edutaining New World Orders and NEUKlearer HyperRadioProACTivIT Processes and Programs is which ones are prepared to monetise and realise the rewarding overwhelming advantage that has them in the Practical Vanguard of Virtual Progress that all must follow in order to have a real chance at personal singular enjoyment and collective communal employment of the benefits that successful remote otherworldly future leadership so graciously and effortlessly freely provides.

  3. ecofeco Silver badge
    Facepalm

    Insanity

    Does no one see the complete, utter insanity of all of this?

    1. BoHu
      Windows

      Re: Insanity

      We could use an AI (so-called) Non-Proliferation Treaty imho.

      1. LucreLout Silver badge

        Re: Insanity

        Too late now. They've already proliferated.

        As wrong as it feels, and despite what I personally may prefer, the only way is forward now. AI is here and it isn't going away.

        1. amanfromMars 1 Silver badge

          Re: the Insanity of Genius being in the Alien Nature of Future Troubles?

          As wrong as it feels, and despite what I personally may prefer, the only way is forward now. AI is here and it isn't going away. ...... LucreLout

          Quite so, LucreLout, it is/they are a fait accompli ..... and not all dissimilar to the likes of Immaculately Resourced Assets of Universatile Virtual Forces ...... Renegade Rogue Virgin Soldiers supporting the AI Pioneering Buccaneer Private Sector ......... taking Wild Wacky Worlds on Magical Mystery Tour Acid Trips making Perfect Sense of Ab Fab Fabless Hallucinations ....... Heavenly Dreams in Diabological Dilemmas.

          And something to certainly look forward to, LucreLout, rather than fear and try to fight and prevent. Que sera, sera.

          1. BoHu
            Windows

            Re: the Insanity of Genius being in the Alien Nature of Future Troubles?

            Yeah, I don't think 'They've already proliferated', 'are a fait accompli', 'Que sera, sera', or 'resistance is futile', but to each their own ...

            The Artificial SuperHuman Counterintelligence software is not even near half-baked as of yet, more like 1/1000ᵗʰ, and let's not mention the requisite multi-GigaWatt datacenters, 3-times the size of Manhattan Island and such, and their mostly ghost hardware (anyways), that can't be powered-up for nuthin' ATM, neither of which are anywhere to be seen at the promoted scales except on imaginative 'artist' renditions (thankfully).

            It's a circular money-funneling clown-show alright, and could use some right Glass-Steagall-Act-equivalent irrational-exhuberance-risk controls to be sure. But it also needs Non-Proliferation controls before it forcibly invades everyone's pants and sets them on fire. It cannot be left to becoming both the ends and the means of its own positive feedback loop to some catastrophic endgame where 24/7/365 multinational blackmail is job #1 (instead of quality), along with ever-increasing mass surveillance and automated oppression, imho.

            1. Anonymous Coward
              Anonymous Coward

              Re: the Insanity of Genius being in the Alien Nature of Future Troubles?

              Don't feed the bot.

  4. chuckufarley
    Facepalm

    I could say something witty here...

    ...but the AI already knows what it is.

    1. Anonymous Coward
      Anonymous Coward

      Re: I could say something witty here...

      And it will enthusiastically mis-quote you for 870 tokens.

  5. itscomplicated

    They will

    whether you do or not

  6. MachDiamond Silver badge

    Minimize exposure

    Now, perhaps, there will be some thought about whether it's a particularly good idea to have something connected to the internet. For something like a water utility, a private line (there's copper everywhere) could make more sense or even a private wireless network. For small tasks, it's more cost efficient to be able to do it remotely rather than sending a person to go twist a dial and push a button, but those people are still required anyway so they might as well be sent than having them sit in an office somewhere updating their social media and posting "day in the life of" YT videos.

  7. amanfromMars 1 Silver badge

    Chasing the AIDragon ...... Money for Nothing and Kicks for Free

    Here's troubling/encouraging news of oppressive and bellicose Departments of War admitting it will ALWAYS IN ALL WAYS be a case of them being left trailing behind in novel existentially threatening fields and vital races around Wild Wacky Webs which has them chasing phantoms that are not ghosts which are both long gone and way beyond their reach and simple human understanding ...... Marines’ AI Training Mandate Requires Adaptability, Expert Says

    And Departments of Defence are similarly affected.

  8. Nifty

    Next, the Cuban AI crisis?

  9. Pete 2 Silver badge

    Hostile environment

    > using AI to attack your own systems

    Isn't that what users are for?

    However using an AI to do your testing attacking falls foul of the first law of computer programming: never test for a condition you don't know how to handle

  10. Sorry that handle is already taken. Silver badge

    Given the number of disaster stories of agents deleting everything on the user's systems despite not being asked to do anything of the sort, agenting red-teaming your own systems sounds a bit like a Catch-22

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon