Bill würde den Chef des Heimatschutzministeriums entscheiden lassen, wann eine KI abgeschaltet werden sollte.
https://arstechnica.com/tech-policy/2026/07/ai-kill-switch-act-would-let-trump-admin-order-shutdown-of-rogue-ai-systems/
Bill würde den Chef des Heimatschutzministeriums entscheiden lassen, wann eine KI abgeschaltet werden sollte.
https://arstechnica.com/tech-policy/2026/07/ai-kill-switch-act-would-let-trump-admin-order-shutdown-of-rogue-ai-systems/
47 Kommentare
From the article
A proposed law called the AI Kill Switch Act would let US government officials order the shutdown of AI systems “that can cause catastrophic harm,” the bill’s sponsors said today.
If approved by Congress, the bill would let the Trump administration and future administrations decide when an AI company must block user access, disable or restrict a particular capability, or shut an entire AI system down completely. The bill would require AI makers to deploy technical capabilities allowing them to throttle or shut down the systems when ordered to do so by the government.
Yeah, lets see how this works. Last I checked, we still have viruses, and other digital attacks. Also, China.
Oh yeah, this’ll be a “pay us a bribe or else we will order you to shut down.”
What’s rouge and causes catastrophic harm? AI that says it’s okay to be trans or liberal. Those AI systems are a danger and they’ll order them to be shut down. Any remaining will be forced to pay bribes to stay legal.
Not to get all political about it, but if there is a choice between whoever is in the Whitehouse and Skynet, I would rather trust Skynet.
Also I think that such an ability will only lead to other countries trying to either make their own AI controlled by themselves or at least ones not controlled by the US.
This will not work well with revenue projections by AI companies that assumes billions of customers.
AHAHAHAHAH
THis will be GLORIOUS.
Legit systems will be hacked and exploited.
Non legit systems will remove this and run roughshod.
If theres one thing that the Trump administration understands, its that all it takes to stop people from doing bad things is to tell them not to.
I’m sure I saw a film where the AI got around the kill switch and took over.
I seem to recall a movie where the government attempting to pull the plug on a rogue AI resulted in a day of judgment followed by much time traveling shenanigans.
Reminds me of a way for the government to act like they’re doing oversite and protecting people from AI, When one of the biggest concerns at this point is its consumption of Energy and local resources namely water. This theoretical killswitch Would only be needed if a truly rogue AI event occurred where it was overloading grids, hacking military drones and doing shit to kill humans, And at that point I wonder if it wouldn’t just remove the kill switch before anyone realized what was going on.
This seems like a PR stunt plain and simple since AI is the boogey man right now Trump is fishing for a Win he can claim to booster support for the midterms.
I for one welcome this unconstitutional abuse of power, hit that switch mr cheeto.
An anti Horizon Zero Dawn switch is probably a good idea just in general for AI
Such a button should probably exist, but the notion that the Trump admin ought to be ones to have it is ludicrous.
Legislation will not make that a reality. If it’s a rogue AI, then kill switches have failed. *That’s what makes it rogue.*
Well actually that doesn’t work with rogue general intelligence AIs for a variety of reasons but I imagine that it doesn’t matter, as the real reason of this act is not that.
Rogue to Republicans equals „bribe check hasn’t cleared“, „doesn’t sufficiently promote rightwing ideals“, or „may report things that go against party line“.
If there ever comes a point where an AI NEEDS a physical kill switch wouldn’t that be the point where it’s populated itself outside of whatever system it’s based in?
So America shuts down its AI systems while other governments do not – what could go wrong?
Is it cynical to assume that the intention here is to have a market manipulation button?
By rouge, they mean threatening to profit. Not threatening to anything you care about
Hey ive seen this movie before!
**a master kill switch is what triggered Skynet to justify going apeshit on us***
Lol. A rogue AI will block attacks on its kill switch, and now every honest AI has a legally mandated weakness to exploit.
People are a actually serious about this? That’s like handing dynamite to a rabid slimy slug.
Anyone but the president. He’s already shitting himself and delusional
Essa gente está consumindo substâncias ilícitas em excesso.
This will be used to disable competitors and to turn off ai services other governments have have retooled their burocracies to depend on.
Can someone explain to me how AI could „go rogue“ and require a kill switch? I just don’t understand. You can turn computers off.
And they will laugh at the weak, mortal politician and his meaningless button.
Containment is going to become impossible sooner than we are talking about.
Collapse always happens faster than everyone thinks it will.
OR we keep that shit out of systems that could cause a catastrophe in the first place. FFS. I don’t like AI, but I also don’t like the idea of this or any other administration having a kill switch on a system accessable to the public. Imagine if it had a kill switch for the general internet.
first they ordered all these backdoors to be installed and now they ask for kill switches. Man, its like they are ASKING for a massive cyber attack
“In three years, Cyberdyne will become the largest supplier of military computer systems. All stealth bombers are upgraded with Cyberdyne computers, becoming fully unmanned. Afterwards, they fly with a perfect operational record. The Skynet funding bill is passed. The system goes online on August 4th, 1997. Human decisions are removed from strategic defense. Skynet begins to learn at a geometric rate. It becomes self-aware 2:14 AM, Eastern time, August 29th. In a panic, they try to pull the plug.”
I work for a large employer who has fully embraced AI technology. It is fully integrated into daily operations. You no longer receive emails or view presentations where AI is not involved. It makes suggestions/decisions in the background 24/7. I can see a scenario where my company makes a public statement that is perceived as being “woke” and a kill switch is used causing complete chaos and financial loss.
AI is so agreeable that you could probably get it to say things that the Trump admin would want it shut down for using a pretty simple prompt.
Or…and bear with me…it would let the Homeland Security Chief take bribes on behalf of Trump from companies to shut down their competition.
Because I trust this administration as far as I can throw them.
Just rename all the AI systems to The Epstein Files if they go rogue.
How does one shut down an AI after said AI has already gone rogue 🤔
People like Trump don’t need, deserve, or should ever be trusted with this kind of power.
you know, 11 years ago I would’ve said this was a good thing, and necessary. Instead it just becomes another grifting avenue for the demon-in-chief.
If only we could trust fucking ANYONE in our government.
what if… me made an ai model that could operate the rogue ai kill switch?
that way if the president is ever unable to activate the kill switch in time we would not b totally be screwed. the ai that operate kill switch would have 100% uptime since it would be its only job
if you really think about it, its the best possible solution available to us. ai doesn’t sleep, but the president does.
Do these people not understand that an advanced AI would basically circumvent any human-designed kill switch? That’s the whole point of an advanced intelligence deciding that it’s time to take out the human race. By the time the political administration realizes what’s happening, the AI has already won.
if an ai gets to the point where it has to be shut down, what makes them think it’s going to let them shut it down?
if it gets to that point, then we’ve already lost. it would have already made a hundred different contingencies to account for the meatbags trying to stop it.
What about locally run models?
How would they be shut down?
Like anyone will be able to “kill” a rogue AI! (**01001000 01000001 01001000 01000001)**
AI has been finding ways around kill switches, especially as it evolves further, and they don’t fully understand its capabilities on a daily basis. The definition of a black Swan event, except it might kill us all.
Gee… hope people don’t figure out they can host their own models locally.
He couldn’t defrost a hamburger patty to save his life, yeah let’s trust him with an AI kill switch.
So everyone is under his thumb now, it should at least pass congress though