Joshua Krook

Karma: 4

Joshua Krook 14 May 2026 11:27 UTC
1 point
0
on: Only Law Can Prevent Extinction
I did an interesting project in uni on whether the state can exist without violence. I discovered there are many tribal communities without violence or prisoners. They use public shaming instead. The problem is scaling this to a large enough scale of modern states.
The other problem is sociopaths and psychopaths. It seems clear that at least in the US, some leaders are severed mentally imbalanced, which means they feel no shame. Without shame, public shaming is meaningless.

Joshua Krook 5 May 2026 7:40 UTC
2 points
0
in reply to: 152334H’s comment on: The Threat of AI Crimes Are Under-Appreciated
I gave the example of insider trading because I think it’s the easiest to imagine. If you get an agent to trade on the market and it does so via gaining data from a company server, then that alone might be enough. Various white collar crimes are quite small in terms of what you actually need to do to commit them.
Fraud is another example that we are seeing already. Because AI hallucinates and makes things up, it’s kind of already vulnerable to fraud scenarios, and in particular misrepresenting the user it works for. So an AI could say their user is a professional (X) when they’re just a student, or something similar. In many jurisdictions claiming a licence you don’t have is already a crime.
So I agree with the previous user that small scale white collar crimes is feasible at a very large scale. The harder question I posed in the post is whether bigger crimes made of many subtasks is possible. I pose it as a provocation, but I’m unsure currently if technical safeguards will be sufficient and simply prevent this. They might, I’m open to that.

Joshua Krook 5 May 2026 7:35 UTC
1 point
0
in reply to: jbash’s comment on: The Threat of AI Crimes Are Under-Appreciated
I didn’t cover negligence in my post but I cover it in the article. The main issue I see is that negligence requires foreseeability of a crime and recklessness. It’s hard to say the common user would foresee a crime, at least currently. They are prompting in a legal direction so they don’t directly foresee illegality. It’s also hard to argue recklessness for AI developers when they’re explicitly putting in safeguards and explicitly state that the AI should not commit crime.
I do think some of our existing laws will catch some situations, and users who give prompts that are borderline suggestive of crimes. You’re right that some users may be reckless, and I particularly think users who jailbreak an AI will definitely be reckless. There’s also good laws on conspiracies and RICO as you mentioned.
My main concern is that AI may have a greater capacity to break down a task into subtasks than a human could, and hire more humans for more innocent subtasks. If a human does know or suspect then obviously they’d be liable, the issue is how small these tasks could be.

The Threat of AI Crimes Are Under-Appreciated

Joshua Krook4 May 2026 11:33 UTC

3 points

8 comments3 min readLW link

Joshua Krook 12 Mar 2026 22:56 UTC
2 points
0
on: The case for AI safety capacity-building work
Do you factor in the opportunity cost of people doing things other than AI safety research? It’s really unclear to me why someone with a decade’s work of feeding the homeless should work on AI safety, nor if they have experience in public health or other worthy causes. It seems strange to ignore experience and say people shouldn’t do what they’re good at and already “optimised” to do that helps society. I’ve met dozens of people who are in this position who seem pressured to work on AI regardless of skillset, experience, background or interest.