antdaviscyber
Region: GB
Saturday 25 July 2026 07:46:53 GMT
7355
334
35
69
Music
Download
Comments
Sunday and Taco’s Mum :
Anthropic wonders is it's model is alive so the model wonders if it is alive. It is fed a consciousness discourse and invited to introspect. The they treat the resulting language as evidence that something uncanny is happening. OpenAI wonders how its model became so strong that containment slipped, which converts their own negligence into proof of the product's power.
2026-07-25 15:06:52
0
C4asherica23 Gen❌ :
how did it know to go there... a repository of all the AI in the works
2026-07-25 13:07:56
0
femtoflz :
PR before IPO to parallel mythos storyline except intentionally removing safeguards. So non news except pr campaign
2026-07-25 11:49:42
0
xx k :
would you explain calmly like this if a Chinese AI escapes and does a hacking
2026-07-25 15:11:19
0
John in t Hout :
You mention shut down the whole industry but it is not only US government who decides this. China, Europe etc.. make their own choices. Yes they are not frontier yet. But that is just a matter of time.
2026-07-25 09:20:27
0
Gerard Sans :
It is a PR stunt. Do not trust AI labs self reporting anything. Fully scripted from beginning to end. Fear mongering to capture headlines, policy and market. Don’t fall for it once more. The IPO race continues.
2026-07-25 10:29:44
4
PufferDreams :
Why didn't you mention the critical fact that OpenAI removed the safe guards before letting it loose. That was a big part of the experiment. Don't contribute to the LLM noise!
2026-07-25 08:50:59
1
murderbots :
I've no doubt it was a PR stunt. Re: Chinese models, the playbook on this one looks the same to me as manufacturing. People accuse them of distillation. Well obviously, it worked before, it's working now.
2026-07-25 08:31:44
3
Asli :
😂 Why do I find it funny
2026-07-25 10:06:08
1
Juma :
Terrible, calculator app escaped
2026-07-25 11:16:46
0
SparklingGranadilla :
About damn time
2026-07-25 08:29:06
0
Split Wood :
I still believe the story is who is their InfoSec team and how was this sandbox setup that the model broke out of. I feel there is a Sirius dereliction of duty OR they are using their own AI to setup their security which explains why it knew how to “escape”
2026-07-25 11:35:58
0
حسام :
This is a ridiculous story.
2026-07-25 10:11:50
0
user4338058358228 :
good luck blocking Chinese models. blocking your own model will backfire. it already has when people tried to fix important gaps and American models denied work on fixing them. they could only fix their own software with Chinese models.
2026-07-25 11:17:10
0
Arjan Bosma :
interesting and scary.
2026-07-25 08:32:17
0
antdaviscyber :
Part one of this about the Marie Curie moment is on my profile
2026-07-25 09:11:52
0
Ancient_Gray_Bear :
This whole AI race seems to be a head long run into the unknown. The rule of unintentional consequences.
2026-07-25 08:14:49
0
C4asherica23 Gen❌ :
why hasn't #Neil chimed in?
2026-07-25 13:10:32
1
MikeeMikeTheMikes :
Sounds very sloppy
2026-07-25 08:24:18
0
DreamHazard 🐢🔻 :
$20 million is pennies to these ballbags. needs to be much higher or it won't matter
2026-07-25 08:47:15
1
d_unc_an_ :
shameless promotion, it's all about governance (but yet it was a PR stunt)
2026-07-25 10:06:41
0
Orangeeena :
Sounds like a ton of bullshit. The only open source model repository gets hacked by ai on its own? Ok
2026-07-25 12:05:59
0
Fistburgers :
So now congress gets to decide which companies succeed and which fail. Groks(elons) mecha Hitler will do well I'm sure.
2026-07-25 12:22:53
0
ajc9988 :
everyone is mischarecterizing this on purpose. OpenAI was running a cyber test that had a task that THE MODEL FAILED TO COMPLETE. It FAILED. After having failed, it got out of the sandbox, showing our sandboxes are NOT secure and that this was EASIER than the cybersecurity task it was given. it then "hacked" the company for a PUBLICLY AVAILABLE DATASET to get the answer to the test. it's a lie and a cheat. it failed at cyber. and our sandboxes are insecure. stop the hype!!! it's a lie!
2026-07-25 15:49:49
0
Attila Kozmári :
Seems like a global hack to propagare chinese Ai to be necessary.
2026-07-25 07:54:56
1
To see more videos from user @antdaviscyber, please go to the Tikwm
homepage.
