Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Disrupting malicious uses of AI by state-affiliated threat actors

openai.com

Post reply on HN
Print view View original

91–94 of 94 posts

Re: Disrupting malicious uses of AI by state-affiliated threat actors

#91

Post by ben_w »

hackerlight wrote: post #51
pjc50 wrote: post #13

Earlier quoted context omitted.

? In what way?

Just wait until an authoritarian state gets its hands on an AGI/ASI that it controls internally. Turning internal oppression into an unbreakable steady-state and using the AI for warfare against external rivals. Techno-optimists are delusional if they don't get spooked by things like this, which will happen if we as a species somehow can't get over authoritarian dictatorships within a few years to a few decades. Whic…

This blog post should be giving you some optimism, that people are aware of the problem and trying to do something about it.

Me, I'm worried by the possibility of a "You Gotta Believe Me" attack[0]: even if done with the best of intentions (and everyone thinks they're on the good team), it risks a memetic monoculture: https://kitsunesoftware.wordpress.com/2019/12/30/memetic-mon...

[0] As fictionalised by Vernor Vinge in Rainbow's End

Re: Disrupting malicious uses of AI by state-affiliated threat actors

#92

Post by euroderf »

mschuster91 wrote: post #75
nonethewiser wrote: post #52

Earlier quoted context omitted.

It's amazing to me how there are really only 4 named countries (China, North Korea, Iran, Russia). I guess it's probably just more political and state sponsored than I would have guessed. Or it's not based on prevalence of attack sources and it's dictated more directly by some US policy? For example, you might also expect India because 1) it's a massive country with many people, 2) it's fairly independent and at leas…

There aren't that many countries in the world with nation-scale hacker groups - the only ones I'd add is Israel with all their involvement into commercial spyware (and Mossad, whose capabilities likely surpass even the NSA), but they're heavily allied with the US.

> (and Mossad, whose capabilities likely surpass even the NSA)

What makes you say that ? Can you point to some interesting discussions / reporting ?

Re: Disrupting malicious uses of AI by state-affiliated threat actors

#93

Post by metricspaces »

mschuster91 wrote: post #75
nonethewiser wrote: post #52

Earlier quoted context omitted.

It's amazing to me how there are really only 4 named countries (China, North Korea, Iran, Russia). I guess it's probably just more political and state sponsored than I would have guessed. Or it's not based on prevalence of attack sources and it's dictated more directly by some US policy? For example, you might also expect India because 1) it's a massive country with many people, 2) it's fairly independent and at leas…

There aren't that many countries in the world with nation-scale hacker groups - the only ones I'd add is Israel with all their involvement into commercial spyware (and Mossad, whose capabilities likely surpass even the NSA), but they're heavily allied with the US.

[dead]

Re: Disrupting malicious uses of AI by state-affiliated threat actors

#94

Post by int_19h »

RecycledEle wrote: post #66

I am concerned that in repressive regimes, the state will control what AI the people have access to. This will make it easy to create music that supports the regime and almost impossible to create music that is critical of the regime. It will be like the pro-state people having assistants to create memes they imagine for them, while the anti-state people are left with paper and crayons, until the AI paper and AI cray…

Oh, don't worry about that. In repressive regimes, people are already denied access to anything useful (like say GPT-4) because of the sanctions.
Post reply on HN