Not in the field only a fan, but I would want the only successful hacks to be the guys in security, like the cyber security testers.
Idk if that is a similar situation here.
2026-09-29 00:43:36
0
Elliott219 :
Good point, though you can argue that, the test thing they do is the reason why we don't see it in the wild and only in the labs. they definitely need to test in the labs but they need true machines that are outside of the internet. and can't somehow back door and get out
2026-09-29 00:23:46
0
Brixton Angel :
Excellent post Nate
2026-09-29 00:36:09
0
I2C_Jason :
Gain of hacking research
2026-09-29 00:48:04
0
Bishop Whitewind :
I read teamed my sandbox but unlike theirs it actually worked?
2026-09-29 00:32:02
0
fectangle :
probably the same philosophy as the Gain-of-function research
2026-09-29 00:08:42
0
The Curious Engine :
I think this is like the little boy who broke a plate. He lies that it only broke in half by accident, then he admits it is in multiple pieces, and finally, under duress, he confesses that he took a hammer and smashed the plate.
2026-09-29 00:20:54
0
lambidu :
does crash tests make cars safier?
it's kinda required evil... better to break into "whatever" in a controlled conditions... that breach can be isolated and later huge leak avoided...
as of me the question is more about how much we can trust AI frontier labs...
2026-09-29 00:35:57
0
The Shrink Whisperer :
The level of shit they’re red teaming today will be available to non state actors outside of the tech industry in six to twelve months. How close do you want to cut it? Ultimately the risks of going too slow or too fast are both potentially catastrophic, but too slow is probably worse. We’ve never seen what these things can do when they are trained and directed to cause as much real damage as possible.
2026-09-29 01:02:51
0
666bbb.0 :
Hack it up ⬆️
2026-09-29 00:36:05
0
James Westcott :
It's the US Military pushing them 🤷🏼♂️
2026-09-29 00:46:56
0
To see more videos from user @nate.b.jones, please go to the Tikwm
homepage.