- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: Jerome Powell and Scott Bessent met with banks to discuss cyber threats from Anthropic
Posted on 4/10/26 at 1:57 pm to ohieaux
Posted on 4/10/26 at 1:57 pm to ohieaux
quote:
This is a big deal and something that could become a strategic weapon if the NSA/CIA/DOD wanted to weaponize this.
You're looking at it wrong though. Those groups already have weaponized this stuff. What anthropic's model does is speeds up the ability to uncover those types of zero-days. That's why industry is on board, like you mentioned. It would massively speed up their ability to detect and mitigate vulnerabilities in their systems.
This is one of those cases where AI is positioned to be the good guy, to assist defenders to rapidly detect and respond to threat actors like nation-state APTs.
0
Posted on 4/10/26 at 2:49 pm to Powerman
Chinese just had largest data breach ever from one of their supercomputers doing all their military R&D.
Im sure ai models had nothing to do with it, and we all found out about Mythos the exact moment it came into existence and it absolutely was not under lock and key around the same amount of time “someone” was fishing around the chinese supercomputer.
What a time to be alive… geez
Im sure ai models had nothing to do with it, and we all found out about Mythos the exact moment it came into existence and it absolutely was not under lock and key around the same amount of time “someone” was fishing around the chinese supercomputer.
What a time to be alive… geez
Posted on 4/10/26 at 2:51 pm to LSUnation78
I don't know if those 2 things are related or not but interesting observation...
Posted on 4/10/26 at 2:53 pm to LSUnation78
I always get a good chuckle when people hyperventilate about how Russia and China are so much better at cyberwarfare than we are.
If they only knew.
If they only knew.
Posted on 4/10/26 at 2:58 pm to Centinel
quote:
I always get a good chuckle when people hyperventilate about how Russia and China are so much better at cyberwarfare than we are.
Do people say this often?
I know there were some concerns about the Chinese allegedly being able to hack our power grid with all the smart devices
Posted on 4/10/26 at 3:06 pm to Powerman
quote:
At least we beat the Chinese to it
I wouldn't be so sure.
Posted on 4/10/26 at 3:10 pm to Powerman
Glad these guys are getting ahead of it.
Let’s just unleash it against China.
Let’s just unleash it against China.
Posted on 4/10/26 at 3:27 pm to Powerman
quote:
I know there were some concerns about the Chinese allegedly being able to hack our power grid with all the smart devices
That's actually fairly common stuff.
I'm talking more esoteric stuff....like exfiltrating data in an airgapped environment by modulating fan speed after you've compromised the system.
high speed = 1
low speed = 0
Then another device is in the room, listening for that particular frequency/noise and recording.
Sure, it might take awhile, but how would you detect that? That example is real btw, and over twenty years old now.
I know that's not really pertinent to the AI discussion here, but more of an example of how the guys who truly know their shite think. Our three letter agencies are full of em.
This post was edited on 4/10/26 at 3:29 pm
Posted on 4/10/26 at 3:39 pm to Centinel
quote:Yep.
That is what should scare the shite out of you. Not what anthropic's new toy is finding.
And finding obscure exploits makes software more secure, not less.
But let’s be honest, the AI companies are heavily invested in scaring the crap of regulators and congress in the hopes of getting regulated.
Posted on 4/10/26 at 3:42 pm to Centinel
quote:There’s a reason the Trump admin reacted like it did when Anthropocene gave them the finger.
ETA: Oh, and there are a bunch of boys in three letter agencies that are NOT happy right now.
Posted on 4/10/26 at 3:46 pm to Powerman
Steve Bannon did a segment on this disturbing bit of news on this morning's War Room Show. AI is either going to be a net benefit for society or it's going to threaten civilization....TBD.
Posted on 4/10/26 at 4:02 pm to Powerman
quote:
The vulnerability allowed an attacker to remotely crash any machine running the operating system just by connecting to it
I'd call that exploiting it
No it’s not. It’s the possibility of exploiting it. The responsible parties found it before it was used by bad actors.
Posted on 4/10/26 at 4:04 pm to Penrod
quote:
The responsible parties found it before it was used by bad actors.
Or used by good actors against bad actors. Depending on your point of view.
Posted on 4/10/26 at 4:36 pm to Powerman
quote:
At least anthropic is being very proactive about trying to find new ways to defend against the monsters we're creating
They are only doing that because their code was leaked
Posted on 4/10/26 at 5:02 pm to Cosmo
Nah, there's a metric shite-ton of money to be made off this type of thing, and anthropic is the leader right now.
Posted on 4/10/26 at 11:31 pm to Cosmo
quote:
They are only doing that because their code was leaked
Their code didn't really leak. The "leak" people are talking about was already available to those who could parse through their minified and obfuscated typescript used to implement the Claude CLI tools. What leaked was the source map which allows you to more easily debug the production packaged code. However all the logic was already there, just harder for a human to parse.
Furthermore, what leaked was a wrapper for interacting with their models not the models itself. What you end up learning are all the techniques they use to brute force their models into do useful programming assistance. I liked the one where they take a response, pass it to a validator, and a retry until valid code is produced. Implementing tools with models has sorta become like knowing how to talk to a woman...gotta use the right words in the right order or you are in trouble.
Page 2 of 2
Popular
Back to top
Follow TigerDroppings for LSU Football News