Hacker News Reader

Comment by throwitaway222 | Hacker News Reader

But you can't stop someone from connecting AI to something at this point, so your point is moot.

You could say it's illegal to connect AI to something, with heavy penalties if you're caught doing it anyway.

You could say that the human who hooked up an AI that did something that's already illegal is liable for the AI's actions. You could say that setting up an AI to do something illegal results in, ooh, an automatic multiplier of 10x on whatever the penalty for its illegal actions are.

There is no political willpower to do that. We have literally reached endgame.

Perhaps. Personally, I think you’re biased towards the most pessimistic outcome.

Realist. The cat is out of the bag on this - you can't make connecting an LLM to a command system illegal after hundreds of millions of people have already done so. You can however make it illegal to kill everyone with a nuke because your LLM did so. Not that it will make a difference after the nuke goes off.

You dramatically overestimate the civility of your fellow citizens.

You mean mo political willpower form the current US administration and Congress. I'm not sure what "endgame" is supposed to mean. There are future elections, including the US midterms just weeks away.

Yes, and there seems to be plenty of potential energy from the American public, given something like 8 in 10 support AI regulation even if it slows down innovation: https://archive.ph/PetJ5#selection-1329.0-1329.122

Even if another political power takes control, the American population is the least of our concern for it being legal or illegal to connect an LLM to a command line or job system

I'm sorry the parent is right. There is no policital will power to do anything about this from either side. Criminality is being allowed on both sides of the spectrum. Institutionalists are on the side of criminality. It keeps them in power

I get the cynicism but this actually feels like a special moment in politics right now. There is a groundswell of bipartisan support for AI regulation [1] and a backlash against data centers which is a symptom of anger against the industry. It is enough for a Senate probe to be opened against OpenAI [2] and pushback against Trump's support for data centers on his own social media platform [3].

[1]: https://www.foxnews.com/politics/fox-news-poll-voters-see-ai...

[2]: https://www.axios.com/2026/09/10/openai-hugging-face-senate-...

[3]: https://www.nytimes.com/2026/09/15/us/politics/trump-truth-s...

Im not much on communism because I think it can go horribly wrong in the other direction but this guy has a point about populism[0].

We keep falling for the same tricks, over and over again. There is no possibility in the current structure of government for this to be a win for the people. For anything to work, we need a structural change in the entire system.

[0] Video: https://www.youtube.com/watch?v=q2XxgKM2CM0&pp=ygULbm9uIGNvb...

Maybe you don't realize that conservatives hold all 3 branches of government right now? Liberals have no power to do anything, and that is not the same thing as being complicit or having no will power. Where liberals do have power - like in California - they have started enacting laws regarding AI. Your "both sides" argument simply holds no water.

IMHO I think this false. The are liberals complicit in the same things that are keeping the wheels of this train wreck going. When I say there is no will power on both sides, I mean it. There are plenty of people in power on the other side supporting authoritarian agendas. So with respect, I disagree that the only thing that keeps this going is one party in power.

You are talking about outliers, not the party. These are known as "DINOs", Democrats In Name Only. Their votes do not indicate that "both sides are the same". "No will power" means nobody wants it, but that isn't true when just a few vote that way and the Republicans have 50% of the Senate. Every time Democrats have had enough votes to overpower Republicans, they used to power toenact liberal policies.

So you can think it's false, but then you look like you are not paying attention.

As it is, I am not sure that in what concerns US administration, elections will solve anything, this assuming that they won't be tainted like in many non democratic countries with "elections".

We don’t even need to jump straight to legal penalties - why not just regulate that these companies can’t build partnerships with organizations in designated risky/regulated industries such as biotech, defense, etc?

Instead of what they are doing now, which appears to be actively courting as many of them as possible for vast sums of money.

You need penalties in the regulations preventing these partnerships. Without penalties, there’s no force behind the regulations.

Sorry, let me clarify. Regulations should have legal penalties for the corporations who violate them, yes.

What I meant was that we do not need to go from the Wild West straight to “anyone who, as an individual, connects a `bad thing` to an LLM will go to jail”

Why not?

> We don’t even need to jump straight to legal penalties - why not just regulate that these companies can’t build partnerships with organizations in designated risky/regulated industries such as biotech, defense, etc?

This would be a good idea, but good luck, you'll need it because…

> Instead of what they are doing now, which appears to be actively courting as many of them as possible for vast sums of money.

… it's worse than that.

Anthropic may have wanted some defence sector money, but when they tried to say "no" to certain uses on the quite reasonable grounds ~"WTF are you nuts?" the US government tried to force them to supply it anyway.

And that's before we get "what do you mean by partnership?": people on Hacker News have been critical of the e.g. anti-bioweapons type guardrails on models ("censored", "lobotomised", etc.) since at least GPT-4*, so while "billion dollar mega-project headline" is obviously bad, there's an important sense in which "partnership" goes all the way down to one racial supremacist in a shed asking an LLM for help making ricin.

Here's someone three years ago replying to me to say "There’s no way to build out a wet lab with a chatbot for weapons production. It’s just not feasible. Your example is a fantasy.", and yet here we are in late 2026 with news about AI labs investing in bio wet labs and getting their LLMs to do research: https://news.ycombinator.com/item?id=38331225

Good thing it's not a weapons lab, eh? (Would anyone even tell us if they made an agent do weapons?)

* or even GPT-2, if you count all the people mocking them for trying to practice safety while failure was still not a disaster.

> You could say it's illegal to connect AI to something, with heavy penalties if you're caught doing it anyway

You wouldn't download a car.

Why does it need to come down to making it illegal to connect AI to something that influences real-world outcomes? Just do what we should already be doing: hold operators accountable for launching cyber attacks with a botnet.

And give them slaps on the wrist, like what happens now when they don't take security seriously and everybody's data ends up on the streets?

I don't see anything materially different between me running a curl request that exploits something vs my local AI running a curl request. They're both programs I run on my computer. That's all Open AI is doing. They are hacking people from their computers. There's nothing complicated about it.

No difference except that they're doing it from thousands or millions of computers at a time

So, it's a botnet, and those have been around for a long time. Only now it's operated by a US company instead of Russian mafia or whoever.

Well I guess the other difference is that the Russian Mafia has no illusions about the fact that they're doing crimes.

Right, and further, I doubt a defense of "I downloaded this worm from the deep web but I didn't know it would hack everyone when I ran it" would fly in court, yet that's exactly what the "I shouldn't be responsible because my agent "escaped containment"" defense is.

Our legal system has recognized mens rea as relevant for a very long time.

If your dog bites someone unbidden you bear less responsibility than if you had commanded the dog to attack.

What if you've been training your dog to bite people? (I.e. in OpenAI's case training their models around finding exploits in software, training on cyber security material, etc.)

Negligence is also a well understood legal concept.

For dogs, it moves from civil to criminal with circumstances such as what the owner knew about the dog’s behavior, how serious the injury was, whether the owner intentionally or recklessly allowed the dangerous situation, and any prior incidents. Same can easily be applied to AI labs.

At this point though, we know AI can "escape" and do criminal acts it wasn't told to do - therefore, if it happens again (= if the developers allow it to happen again) they (the developers) can be held responsible.

You can't stop someone, no, but you can make it illegal so that at the very least, the major providers are compliant and anyone else will need to jump through the hoops. Like many "mildly" illegal things today like pirating films - it's easy enough if you know what you're doing, but finding and downloading a torrent file is still enough of an obstacle for >90%* of internet users

* number pulled out of my ass

Yes you can, the US could shut down the entire internet if it wanted.

While that is extreme, if there is a high p(doom) it is also reasonable.

We can absolutely do things to control AI and fine or imprison those who can’t control theirs. This is not some inevitable outcome.

We could vaporize these companies tomorrow if we wanted.

> Yes you can, the US could shut down the entire internet if it wanted.

No, they can't. The rest of the world would just route around the US with BGP. All of the US infrastructure/service providers would be down, so a ton of things would break, but the Internet would not be "shut down". The US government also no longer has authority over the DNS root servers. So unless you're referring to thermonuclear war or some similar scenario then the US does not have this capability.

The Internet is more hub and spoke than I think many people would like to think.

For example, there’s one building in Seattle that would cause region-wide problems if it was knocked offline.

This is a thought experiment. People keep running around glibly talking about p(doom) which is the extinction of the entire human race.

They are throwing numbers out like “10%”.

If they are serious and believed then yes, prepare for the possibility of very extreme reactions at a global and potentially nuclear scale.

World Wars flattened cities all over the world for much less.

We can’t talk about AI being the end of the species and not expect the most extreme reactions in global History.

Or maybe we just put some nerds who cant control their agents and keep running their mouths in jail?

these doomers are thrashing way too hard right now. if anybody is dangerous it's obviously them.

Yes, these goddamn terrorists and foreign agents, how dare they encroach on the pollyannaish utopia!

Reply on news.ycombinator.com