Forgot your password?
Close
typodupeerror
AI Government

AI Execs Game Out 'Political Revolt' after a Catastrophic AI Event (yahoo.com) 47

Posted by EditorDavid from the what-if dept.
Top executives at Anthropic, OpenAI and other AI companies "are privately gaming out scenarios for a public and political revolt after a catastrophic AI event," reports Axios, one that turns "an already wary public further against the technology and its leaders..."

These AI company executives "anticipate a large-scale event, most likely a cyberattack, that shuts down access to financial services, internet connectivity, or even power and water..." the article points out. "Many top AI researchers and executives believe a major incident is inevitable. "OpenAI conducts preparedness exercises where teams discuss and work through a range of potential scenarios," a company spokesperson said. "These scenarios are not treated as inevitable, but are meant to help us prepare for a variety of circumstances." Anthropic declined to comment...

Top AI planners assume Democrats, ascendant after the midterms, will move fast to shut down AI but will face significant challenges... The planning involves red-teaming for worst-case scenarios. It focuses mainly on racing to educate members of Congress. Company executives know regulation has no chance of passing right now, but still want to shape the legislation and policies U.S. leaders will turn to after a first catastrophic event....

Many AI industry insiders told Axios they believe a major event will occur in the next six to 12 months.
"Even with Democrats in control of the House and Senate, the companies see fractured politics as the norm, including within the Democratic Party."

AI Execs Game Out 'Political Revolt' after a Catastrophic AI Event More | Reply

AI Execs Game Out 'Political Revolt' after a Catastrophic AI Event

Comments Filter:

  • inevitable is what they want (Score:4, Interesting)

    by dfghjk ( 711126 ) on Saturday October 10, 2026 @08:35PM (#66370360)

    "These scenarios are not treated as inevitable..."

    But creating agents that enable those possibilities IS treated as inevitable, because AI companies want the catastrophes to be enabled for their profit. The ability to do disastrous things is the plan, so the disaster is inevitable.

  • I get that after the torment nexus goes off everyone will hate you and battering ram down the doors of your bunker full of child sex slaves, but like, what if you just didn't build the torment nexus to begin with?

  • But to really fuck shit up, you need a machine.

  • Is Altman in jail yet? (Score:1)

    by Anonymous Coward

    Hacking Australia's government websites...

    Incarcerate OpenAI.

  • Not the brightest bulbs (Score:3)

    by Berkyjay ( 1225604 ) on Saturday October 10, 2026 @09:16PM (#66370390)

    These are the same people who think their software is sentient right?

    • That was just that one guy. He doesn't work there anymore. But in retrospect, that episode was before ChatGTP made us all aware of how far AI has gotten. It's more understandable in light of that.

      • Anthropic's entire executive team thinks this and has been lobbying religions leaders on the subject. They've also started banning people from being abusive to Claude.

        • They've also started banning people from being abusive to Claude.

          Now, I can't get the image of Claude wearing a top hat and tails singing "putting on the ritz". For the love of God don't humiliate him.

    • These are the same people who think their software is sentient right?

      Roughly, yes. This is significantly coming out of the “Effective Altruism” movement that is especially concentrated in Anthropic and that is trying to lead the creation of top down government regulatory bodies. Of course the whole EA philosophy isn’t Anthropic’s “official” position but the strong influence is undeniable.

      Concerned folks, such as the entire All-In crew despite their internal political differences, have said that this is a self fulfilling prophecy. They poin

  • "Even with Democrats in control of the House and Senate, the companies see fractured politics as the norm, including within the Democratic Party."

    I don't need a frontier LLM to tell me that the Democrats are ineffectual.

    The problem with actively trying to be centrist to please everyone is that a) that's impossible and b) it means actively trying to find candidates who are less progressive than the one you're looking at when you're also supposed to be the more progressive party.

    • The counterpoint is that America will always swing back from wherever you try to push it. Lately the magnitude of the swings has been getting larger and sooner or later that will be a problem. Calming things down with a centrist should give us a more stable country......IF the other side recognizes it as such. But if they try to paint the centrist as an extremist as they did with Obama; then you're screwed anyway.

      • Obama was not a centrist. He gave us a huge socialist edifice in the ACA and was remarkably cavalier about bombing other nations, even more so than Trump or Bush by total nations. He was excellent at removing illegals though, so maybe kinda a centrist? Not sure those balance out.

        • Of all the ways they could have done a healthcare bill, they chose options that were centrist over those that were not. Rather than do single payer or the public option, they picked the individual mandate, a concept originally proposed by the conservative Heritage Foundation in 1989. Many of the concepts in the ACA were modeled on Romneycare. Looked at objectively, it's hard to claim the ACA isn't centrist.

          If you still doubt it, watch any of Obama's SoU speeches. Appealing to both sides was important to hi

          • Real reform would have meant reforming the way insurance had become a bureaucratic nightmare at all levels. Obama was too cowardly to fight the insurance lobby (and to be fair, so were his predecessors on both sides), so instead he made the problem worse. ACA has raised everyone's premiums by more than triple over 15 years, with worse patient outcomes.

            • There is a provision in there restricting how much of insurance premiums can go to overhead; something like 80% must go to patient care. Assuming that's being enforced, the problem isn't entirely there. (Not trying to defend the insurance industry.)

              Increase coverage for the sickest was inevitably going to increase costs. The intent was for wider buy-in to counter that. But when the individual mandate was struck down, that went away. It should be no surprise prices have climbed since.

              • There is a provision in there restricting how much of insurance premiums can go to overhead; something like 80% must go to patient care. Assuming that's being enforced, the problem isn't entirely there.

                No, this is the exact reason why the ACA is a disaster. Insurance company profits are capped at a percentage of the cost of care, so they are motivated to drive up costs. But what about the uncovered treatments that the hospitals have to eat?

                • Ah, that's an interesting way of looking at it. I do wish we could have devoted some effort to fixing the thing rather than going all in on repeal. There's a lot of useful stuff in there alongside the bits that aren't working.

                • No, this is the exact reason why the ACA is a disaster. Insurance company profits are capped at a percentage of the cost of care, so they are motivated to drive up costs. But what about the uncovered treatments that the hospitals have to eat?

                  I heard exactly this during a consulting project in the health insurance space. We set up the ACA so that driving up costs made sense to hospitals and insurance companies who manage the major entitlements (medicare, medicaid). Zero incentives to drive costs down.

        • As a dual American/Canadian citizen I can say that Obama's ACA was not a huge socialist edifice. Rather it was an insurance-industry-friendly alternative to single payer (European-style "socialized") medical systems where citizens can expect much more efficient care, funded by everyone's taxes at far lower cost

          Having grown up with socialized healthcare as a US military brat, I can say that I am more comfortable having my children covered for care with no threat of bankruptcy here in Canada, than I would be

  • It would be negligent of any company not to have a plan for any foreseeable adverse result of their product.

    • Seems like everyone missed the big joke, though... they used their AI to run these scenarios, so now they just trained their AI to carry out those scenarios. Even if the machine they ran the scenarios on was airgapped and they took the WiFi card out and completely isolated it at the time, if that machine ever is connected, then it's possible that the scenarios could be absorbed by the main AI.

  • ... such an event is actually possible. Far too much critical infrastructure is not even remotely secured as well as it should be. Hackers usually do not go after it, because there is no profit in it and scary people may come looking for them. But AI, while not very competent as attacker, may not understand that and may attack anyways.

    • Every infectious disease must find the right balance between aggressiveness (which kills the infected so fast that the disease can't spread) and self-limitation (which permits infections to spread undetected).

      It's a matter of life and death for any anthropomorphized plague.

      • I think your comparison to disease is apt. Consider the recent Russian who died. Now consider how we treat infectious diseases. We do not allow them to interact with the general public at large. We quarantine to a very high level. And when there is a failure, all countries take a keen interest in the how did it happen and how can we prevent it from happening again. Now look at AI, where they can't even bother to keep it disconnected from the web during "research". Really, the answer is jail for the research
        • There definitely needs to be *legal* push-back against this "rogue AI" bullshit.
          You ran an LLM with a set of tools that it used to break the law, or cause damages to another party.
          The key word is "You". There is no "rogue AI", there's you running metasploit against a target and trying to say a rogue metasploit did it.

          Jail is a bit harsh- but definitely standard legal treatment of them.
          I was busted after breaking into AT&T when I was 17. I didn't go to jail. Definitely had the fear of god put into m
          • You got lucky. A person I was in a HS class did a prank in college that was a break in to get something for a frat. He did get jail. I expect that 17 saved you. Barely. But I think jail is required here. Nothing less is going to stop this insanity. I expect fear of jail would be a very good deterrent for the cohort doing ai. All of these things are text book criminal, breaking in for monetary gain. My HS mate got jail and I doubt it even qualified as monetary gain.

            I saw another story about booz allen runnin

            • Jail- wild.
              mate- British? Aussie? Perhaps laws were more severe over there at the time.

              In general, in US jurisdictions, this would fall under negligence or recklessness, as the researcher likely didn't have intent to break into the remote system (and proving it beyond a reasonable doubt would be pretty difficult anyway)
              Recklessness has a decent probability of leading to jail time depending on the severity of the fallout. Second time you do it is usually a year behind bars if your first sentence is suspe
              • US, Indiana. Note you were 17. Below the special age of 18. He was in college, so above 18. Frat rush process, so sophomore year, always 18 by then unless you skipped grades. Again, you got lucky, you should be very thankful.

                Certainly at this moment, reckless, not negligence. They are researchers in ai doing this. I'm not a researcher in ai and even I know the stories. Frankly I'm amazed that no bar organization has dis-bar'ed any lawyers for submitting ai hallucinated cases.Some judges have done summary ju

                • Again, you got lucky, you should be very thankful.

                  Na, I think there is probably just a stark difference in enforcement goals between WA and IN.
                  Throwing a first-time offender in jail for a crime that caused no damages is abnormal- or at least is back in the late 90s, here. But then again, didn't they toss people into jail for decades there for having a dime on them in IN? Or is ultra-strict drug enforcement in the midwest a myth?

                  I explained why you can't apply them directly- they require intent.
                  An escaped program is not intent. It's negligence or reckle

                  • Been a very long time, but I think there were "damages". And your, "If, for example, some jackass runs a .sh file they found online, ..." example fails to acknowledge we are not talking about a random person. We are talking about an expert. They should know better. I'll toss another whatif. Whatif a bio-researcher allows the plague out of the lab? Is that negligence? Maybe they had no intent, just careless or maybe they didn't RTFM about proper precautions. I'm sticking with they knew, and because they knew
    • This is wrong on multiple points.
      1) Foreign adversaries are hitting infrastructural targets daily. In 2021, attackers attempted to increase lye content in a Florida water supply to lethal amounts.
      2) They're highly competent as attackers. This is demonstrated ad nauseum. I wouldn't say they're more intelligent than a human at it, but they're dozens of times faster, and persistent, and smart enough to find the holes.

      The cognitive dissonance you're suffering must be physically fucking painful.

      • Re: (Score:2)

        by gweihir ( 88907 )

        Wow, you really have no clue.
        1. Hitting and hitting large scale are two different things.
        2. No, they are not "highly competent". Stop pushing that idiotic claim.

  • "This massive and unpredictable event, which will happen in the last week of April (just in time for May Day), will be unpredictable and massive, and will affect everyone except for certain sympathetic members of the press and the political party we support. And it will be massive and unpredictable. On April 29 next year."

  • Colossus: the Forbin Project (Score:3)

    by jddj ( 1085169 ) on Sunday October 11, 2026 @12:54AM (#66370512) Journal

    Old movie from the â60s.

    Bland compared to 'leet haxor looks at Python code on a green-screen and breaks into the Pentagon while the timer on the bomb ticks down to 1'.

    But required viewing, due to its prescience, and current applicability.

  • AI is and will continute to be increasingly empowered with indifferent safety precautions because the power money confers is all that matters to techbros.

    First world systems are far more fragile than they appear to humans. AI doesn't have that problem.

  • Scoop: AI companies plot "the day after" [axios.com]

    “Top executives at Anthropic, OpenAI and other AI companies are privately gaming out scenarios for a public and political revolt after a catastrophic AI event.”

  • Sounds like a good reason to vote Democrat (Score:3)

    by SpinyNorman ( 33776 ) on Sunday October 11, 2026 @09:23AM (#66370734)

    The current situation with the Republicans in power and willing to support literally anything that Trump does, is not healthy.

    We've got the AI companies themselves screaming out to be regulated due to the danger, and Trump instead refusing to do so and being accelerationist.

    Trump is becoming increasingly unhinged, and not just in a har har look at the idiot kind of a way. Yesterday he decided to stab Ukraine (& Europe) in the back to buy diesel from Russia, when he could instead have bought it from a friendly country like Canada, or just have stopped US exports. When your commander in chief is providing financial aid to your enemy, then things have crossed a line ...

    • The current situation with the Republicans in power and willing to support literally anything that Trump does, is not healthy.

      For the ultra-wealthy, chaos is great. They will ride it out with their stockpiled supplies and remote resorts and child rape islands, and if necessary, their bunkers. As we lose our shit they will buy it up and we will own even closer to nothing. They don't actually believe that the USA can fall, which is stupid but predictable, and it probably won't any time soon.

      Trump is becoming increasingly unhinged, and not just in a har har look at the idiot kind of a way. Yesterday he decided to stab Ukraine (& Europe) in the back to buy diesel from Russia, when he could instead have bought it from a friendly country like Canada

      In fact we do buy it from Canada, we buy way more than we just got from Russia all the time. This is just a way acceptable to his base for him t

  • It already happened. Repercussions incoming.

    Status:
    Technological Singularity- IN PROGRESS
    Societal Singularity - NEARING COMPLETION

Slashdot Top Deals

"Anyone attempting to generate random numbers by deterministic means is, of course, living in a state of sin." -- John Von Neumann

Close