So, this is going to continue to escalate, until whoever gave the text generator the software tools needed to do this and gave it the access without any kind of firewall, gets charged with the appropriate felonies and is put in jail.

To be clear, this is illegal activity, and the LLM can only do this if a human provides the means. That human == criminal scum.

Don’t forget whoever instructed it to preform the action.

An LLM can’t just unprompted get the idea to do this.

I mean, hallucinations are possible, especially if you have long-running agents. However, I also suspect direct instruction, simply as the alternative would take too long, be excessively expensive and the “newsworthy” result subject to chance.

Can’t build a PR strategy on that.

Yeah,all of these news articles about AI going “rouge” and attacking other companies and even the government happening at roughly the same time, without any real evidence to show what actually happened to trigger the events…

I am hammering X like I was performing the intro to the song Them Kinda Monkeys Can’t Swing by Slade, as I keep reading about this crap.

For reference, here is the song mentioned above:

youtu.be/Ojk8gAEdat4

They’re trained off masses of shit with no regulation. For all we know the idea is from a mash of 2011 Reddit comments and an 2016 news article about public health in The Netherlands.

LLM control is just trial and error guardrails applied like band-aids, many of them contradicting each other.

This is what happens when technology is rushed to implementation before governments can react with regulation.

These are USAIaian businesses attacking other countries. Why would anyone get arrested?

So, this is going to continue to escalate, until whoever gave the text generator the software tools needed to do this and gave it the access without any kind of firewall, gets charged with the appropriate felonies and is put in jail.

there’s hacking nearly every week and no ones in jail. Medibank, Qantas, Optus blah blah blah These arent climate protestors, they’ll not see jail time.

Why do I feel like the word “known” is doing some heavy lifting here?

So who is going to jail now? I am sure a company is responsible for its actions, right?

The government can choose to sue (civil) or prosecute individuals/executives (criminal liability).

Assuming there is some legitimate explanation (I’ll be interested in hearing this one!) they likely won’t, but will now establish higher level reporting channels with OpenAI and Anthropic, and perhaps check their important email boxes more often regardless. ;-)

This is a big problem with AI, laws have not remotely caught up to these scenarios. Then you have Argentina creating legal structures for AI corporations that could essentially do this type of stuff, but evade prosecution because the corporation will give the owner limited liability and with no employees, the only punishment can be to shut down the corporation. Then they can spin up a new AI corporation and do it all over again in theory. Technology is advancing far beyond our ability to control it properly and tech bros don’t care much about consequences since they have historically been so minimal. Move fast, break shit is a psychopathic ethos.

Its easy. There are laws against hacking. If you used a script its still you who go in jail. If you used an Ai its still you who go to jail.

Do not let them try to spin this into the autonomy narrative.

Someone set up the agent, somebody gave it the prompt that caused this to happen. Some one did not do their job properly.

Ai agents are not independent actors, they are tools.

If I missmanage a forklift, I am held responsible. Breaks failed? I failed to do the right maintenance. Load fell down? I did not secure it right.

This is not some impossible gordian knot to solve…

I’m not saying I agree with the narrative. I’m saying it will be attempted and will be a problem before the legal system can catch up and address it. We will also be talking international law, which is trickier, as countries jockey to make friendly laws to get AI cash like Argentina. Especially with the tech industry constantly suing and lobbying.

I strongly disagree that the tech is too advanced to control.

It would be trivial to monitor the inbound and outbound traffic from these models, but they don’t bother.

If I shoot you with a gun it’s murder. If I shoot you with a robot-gun it’s equally murder. There is no problem with the law, only with enforcement.

I think there is enough here for a judge to order a search warrant. It doesn’t mean yet a crime has been committed, but there is enough to start some kind of compelled discovery process to find out what the fuck is going on.

Since it’s an important enough corporation that got caught, in a sane society it should not make a difference, but we low key have a two-tiered justice system from the looks of it.

Australia gave up on their sovereignty when they chose US submarines. They’re just reaping what was sown at this point (not submarines, that’s for sure).

And then…?

What?

Still, nothing?

The breach happened in June but Open AI only became aware of it in August. It notified the government via an email, external to a general inbox on 10 September

This is just atrocious on every level. First of all, they’re hacking everyone with impunity. Then, they’re taking 2 months to find out about it, and a further month to report it. So not only can they not control their AI, as a company they clearly have no systems in place to report it either.

They absolutely can control their agents, they are just choosing not to or are just blindingly incompetent.

I going with “all of the above” on this one.

Reading the story it’s not clear who set up the agent. It’s an agent running on OpenAI but it’s not clear whether they’re saying a customer instructed an agent to hack the Australian medical board and got around guardrails to prevent that, or that they’re saying their own internal work agent decided to hack the Australian medical board at random.

In both cases OpenAI has culpability. In the latter case, they should be held fully responsible.

My understanding is that it was the latter. From the ABC (Australia):

How was the data accessed?

As the government understands it, in this case, OpenAI gave an agent a “benign” task of trying to do research about public medicines spending.

The model then searched the internet widely for the information, and came across the Services Australia portal.

It then asked questions of the portal, but the portal didn’t provide it with the requested information. As a result, it gained unauthorised access and secured information that wasn’t public.

There we go. Fully their own responsibility, then.

It would be AWFUL if someone used AI to hack the Department of Home Affairs and grant all 417 visa applications. Just saying.

Edit: Maybe only the applications that are 100% decision ready and have been lodged on the 5th of July? No? Ok.

Hold on, they might be onto something. (Not this target, obviously.)

They use AI to hack into governments, they say it was AI agent or something I didn’t know.
They bomb a school in Iran, they say it was AI.
Sure, let’s throw it to AI! Not USA who does war crimes.

Can I sue AI because of its “errors”?

No of course not its a tool you can’t sue a tool.

Oh so I can sue the user of the tool. The one using it is responsible?

Naaahhh it was AI it all did it by itself we can’t do anything about it.

Its Anthropic OpenAI. Didnt notify government for more than three weeks. No apology. Meanwhile Albanese and Labor party to water down copyylaws so AI can pillage Australia content creators for no money.

Anthropic hacked places using an openai agent??

Albo has got his head up his ass playing centrist.

Vote Greens.

accountability is needed here; heads need to roll

Now watch, in the US, an agent will infiltrate the election process somehow in November which will lead the government (Trump Mafia) to declare the election invalid despite the fact that the democrats won. I’m just waiting for the headlines.

This is US state sponsored.

Awesome. More, I want AI to hack way more government shit everywhere. Fucking love it.

It’s not an agent, it’s whoever used whatever piece of software in whatever which way to connect to those systems, that’s who is responsible and both them and the people who set up an insecure Medicare network should be in jail. For a long time.

There is a company called CompuGroup Medical which is all-in on “AI” (LLM) and with a CEO funding far-right propaganda portals (Nius).