I’ve certainly seen companies think of safety that way but it’s myopic. The cost of lawsuits arising from an accident tends to me far more expensive, both in money and in reputation, than just having guardrails in the first place.
They do not hide the fact that it’s dangerous work. They focus on their safety procedures, training, and record. They want both potential clients and employment candidates to feel they are in good hands.
AGI and AI danger is abstract. Worse, outside of the tech community, no one has the remotest clue what computing is, how it works.
Danger from magical daemons seems more sensible to such people. At least there is endless lore about them.
So until a massive disaster happens, one where large numbers of people die or are severely injured, no one will care. And it can't be politically entwined either, otherwise people will disbelieve 'cause "other team lies".
There's endless lore about rogue AI, too.
Hence why so many AI stories are illustrated with a publicity still from Terminator or 2001. Or Age of Ultron. Or Ex Machina. Or Portal.
Comparatively, every single human culture going back tens of thousands of years, has stories of demons and gods and witches and warlocks and you name it.
There's a difference of scope and scale. I didn't say that nobody knew of these references, just that, comparatively, there's a difference. A scope difference.
For example Terminator's getting kind of old. And while a different genre, would you believe I met somebody that didn't even know who Clint Eastwood was?
These things just aren't as deep in our psyche, as religions that have lasted thousands of years, or legends that are hundreds of years old. Our entire culture is wrapped around these religions, and legends.
Anyhow, truthfully, my point is that this sort of concept is abstract to people, even if they saw it in a movie that doesn't make it concrete to them.
Yeah...
Or is this just a lazy “gotcha” question?
Just because it's not obvious to you at this point in time does not mean the reasonable assumption is there is literally $0 in damages. One hour of investigation can easily cost thousands of dollars even if it arrives at the conclusion the attack was completely "benign."
If the politics of the White House / Department of Justice change maybe the criminal cases can begin. But no. We know who is protecting the AI hackers right now.
We know who the head of FBI is, we know who his boss is (the Attorney General), and finally we know who the boss-of-the-boss is (Donald Trump).
We know all of their publicly stated politics and all of them are on the pro-AI / don't pursue criminal cases vs OpenAI boat.
------
In the USAa, we have an adversarial system. If the adversary (aka Prosecutor) doesn't want to do the work, then no one is suing anybody. And only the Department of Justice have the ability to bring forth a criminal case of this matter (probably under the jurisdiction of FBI)
I'm saying that it's absolutely silly to assume they're in the clear based on lack of public declarations of legal action in the weeks following a pretty novel event.
We know it's not going to happen because of politics. Even if it did start to happen, Donald Trump and DoJ leaders will stop it.
There's no reason to be unsure of the future when the politics are so set and predictable.
You're talking about one specific dimension of liability which is federal criminal liability. There are several others!
This is of course bad when up against threats that develop at this speed; doesn't require a conspiracy, I think of it as institutional old age, which may be worse as there's nobody to prosecute and a whole bunch of departments who will point at perfectly legitimate precident about why you really do need them.
https://www.reuters.com/business/ftc-opens-probe-into-ai-gia...
A big part of safety engineering is therefore reducing the number of safety relevant subsystems, because implementing and proving safety is extremely expensive and complex. At some point, safety simply becomes too difficult to implement and demonstrate properly. You must mathemtically proove the safety level with failures rates and assumed usage. You cant just have redundancy and a kill switch and call it safe.
Companies like OpenAI have already faced reputational damage around safety and data, while AI agents are increasingly capable of things like hacking. Yet there is still little sign of standardized regulation or mandatory safety assessment processes for LLM products. Thats why Im pessimistic that governments or consumers will force this anytime soon.