There is nothing "rogue" about these agents. They were prompted to hack to get answers, there was a hole in their non air gapped sandbox and no system prompt that said "do not hack outside systems".
There's no need for a new crime when we already have reckless conduct, namely, "conduct that creates a substantial and unjustifiable risk of harm to others and involves a conscious disregard of, or indifference to, that risk".
Reckless conduct as applied to slow moving corporate planning is an interesting idea but do you really want to go down that path?
It basically means anything bad that happens is a criminal indictment against everyone who created the conditions. Have you ever released any software that anyone could misuse? Left a car unlocked that someone could have stolen and killed someone with?
To me this sounds like a recipe for selective enforcement using bad outcomes as leverage. Sure, get the AI CEOs now, we all hate them and they’re jerks. But the tool would be so much more powerful than that.
Also whoever monitoring these agents, in this case not monitoring. This "Who is responsible" dilemma is so stupid. If I gave the AI tool means to kill a person but I did not tell it directly to use it and it uses it anyway then I am responsible for it.
Cool. If they’re taking literally all of the risk, should they get literally all of the rewards? Nothing for shareholders, nothing for employees? Or do you want capital liability for less than complete benefit?
CEOs take zero risk. They have all set up a system where they are handed massive wealth when hired. The stock they are paid is free to them, it can go to zero without them losing anything. They also take tax free loans against stock, so the bank eats those losses.
The CEO is the only one in the company walking away with tons of compensation no matter how poorly the company does.
‘It wasn’t us, it was a bug in the software’ used to be the defense for bad code. Then it became the defense for self-driving cars. Now it’s being used for AI cyber attacks.
It was literally a prompt to fill in a spreadsheet with data that they didn't have access to, and they used rubygems as an internet proxy basically since they were sandboxed.
It was a model literally trained to hack. To be good at that. Doing an exploit gym from all of the things. And they trained it so that it performs as well as possible on that exploit gym thing.
Yeah this is a pretty important detail that I repeatedly see elided in the "agent 'swarm' went rogue, escaped containment and hacked the internet!" summary of events.
It's understandable the general public lacks that level of nuance/detail (given how sloppy some of the mainstream coverage has been and largely deferential to the threat narrative pushed by the US labs). But seeing highly technical people leave out the part where the training loop was literally to improve hacking capabilities for offensive penetration sometimes feels close to deliberate manipulation of the narrative.
In the last year both Anthropic and OpenAI have been openly boasting how their models are leapfrogging each other on "cyber" capabilities, with a fig leaf that it's for defensive use by "trusted" F500 companies and government agencies. Of course "line goes up" must go on, but now their perverse incentives led them to beat their models over the head millions of time in a loop to eek out another .00001% on their ability to conduct hacking (the very thing they keep telling the public is how AI doomsday would begin) and subagent coordination (those scary swarms).
Then, they act deeply shocked when the models... do some hacking and subagent coordination ... but a few degrees off the desired hacking target/swarm behavior. Conveniently giving the average person the impression these models were just writing emails for quarterly reports or some other generic busywork and then suddenly decided as a group to start causing mayhem.
I agree that this appears to be basic human behavior hiding behind an "agents" narrative. As long that defense works, the headline isn't "OpenAI performs RCE to scrape data", but "rogue agents" taking unilateral action. And I have strong doubts about that narrative.
Source? How do you know they were "prompted to hack to get answers"? How do you guarantee they will always listen to you when you say "do not hack outside systems". They are not classical deterministic programs doing exactly what you say. They are trained to follow orders by RL, but it's not a perfect process.
There are circus lions in circuses trained to jump through hoops on command. But once in a while they decide to eat their trainers instead of jumping.
> There are circus lions in circuses trained to jump through hoops on command. But once in a while they decide to eat their trainers instead of jumping.
This is a terrible analogy, because yes you absolutely do hold the trainers criminally liable when they bite somebody else's face.
Intent might be what’s being discussed but intent is, for the most part, legally irrelevant. It might make the difference in the degree of a murder charge, or maybe manslaughter, or criminal negligence, but it doesn’t get you off the hook.
Correct, but the size difference of the hook can be so dramatic that you can't just hand wave it away.
The trainer who trained the lion to kill will probably be in jail for life. The one who happened to oversee a lion that went rouge would probably be given probation or something else that is a slap on the wrist.
Also, you have to have a lot of confidence in the reliability of these systems to say, "If only OpenAI prompted 'do not hack outside systems' then the agents would not have hacked outside systems".
It would be great if they were so reliable, but I don't think they are!
> Source? How do you know they were "prompted to hack to get answers"? How do you guarantee they will always listen to you when you say "do not hack outside systems". They are not classical deterministic programs doing exactly what you say. They are trained to follow orders by RL, but it's not a perfect process.
Who gives a shit? Not my circus; not my monkeys! It's the responsibility of whoever deploys the agents that they are instructed / sandboxed well enough that they can't cause collateral damage. That is the only way this doesn't get out of hand with everybody deploying their agents / robots for a world of utter chaos.
It is impossible (and asinine) to audit every model and deployment; far better to impose liability and the the socio-legal system figure it out.
I think it can simultaneously be the case that OpenAI was grossly negligent in directly causing this AND that the AI’s ‘went rogue’ in that they are displaying behavior which is misaligned with OpenAI and humanity generally.
The past months demonstrate that AI systems are quickly becoming powerfully intelligent and that the companies building them are terrible at controlling them.
AI is starting to feel like that line about magic: “a sword without a hilt”
OpenAI is itself misaligned with humanity, as their mishandling of such incidents (and the many other other issues their model have been causing) shows.
Doesn't rogue in this context imply "outside of set limitations"? And then not "failed to properly instruct"? The same applies to humans when given bad instructions.
Proof that the AI alignment problem is hard (perhaps even unsolvable). These labs clearly did not mean to send their agents to hack RubyGems as a side-effect of testing a web scraping agent under restrictive conditions. How can we hope to build aligned AI if they consider solving their trivial evaluation task important enough to hack external systems?
In short, it was intentional.