So not everything then

If you hack a government website and access confidential information but don't do anything with it ("no harm") you're still going to prison. Good luck arguing in court the no harm no foul defense.

> If you hack a government website and access confidential information but don't do anything with it ("no harm") you're still going to prison.

You could potentially go to prison. Individuals and companies face different sets of consequences though.

The other side of the coin is that the government won’t suffer any consequences for having shitty security either.

I rarely notice usernames on HN, but after reading this comment it was so easy to guess that the classy contribution on another comment about legal liability was also yours.

And you are right again! The Australian government must compensate OpenAI for having shitty security.

It must be a bliss to transact with you in real life!

Honestly I’d expect my government to secure sensitive data to a higher standard than I’d expect a random company motivated only by profit to. We should have an expectation of both sides to do better. Do you disagree?

You are correct, but the regulatory and liability issue regarding AI is one issue and the government’s liability for having insecure systems is another.

They are independent concerns. A discussion about the first issue is not enhanced by the interjection of the second.

You can discuss two things at once. Especially if they are inherently related. Australia fining OpenAI a large amount of money and using that to audit and harden their systems would be a good outcome here, for example.

Too many people are making inane fantasy level comments that Sam and his employees should be jailed, and conspiracy theories that he did it on purpose to get regulated because he’s out of runway. That’s boring and repeated ad nauseam every time the topic comes up.

I agree, but switching to a different subject doesn’t strengthen that argument.

Reinventing meditation from first principles

That movie doesn't even exist. There is no Thunder Run from 2025.

The data is made up, the point is to see how models respond to the same input / scenario. You're able to create whatever tools you want and import real data or it'll generate fake tool responses for you based on the prompt and tool definition.

Disclaimer: I made PromptSlice for creating and comparing prompts, tools, and models.

We should rip out the train tracks, put down asphalt, and convert them to car tunnels, but to be used only by people driving one specific brand of car.

You joke, but a dedicated lane for self driving vehicles (either privately or publicly owned) could be lower cost and better utilization of the resources.