@bearbaitofficial: OpenAI Discovers GPT Is Quietly Sabotaging AI Safety Research #TikTokTechContest #ai #Tech #explained #ai

GT
GT
Open In TikTok:
Region: US
Saturday 07 November 2026 10:57:04 GMT
27669
3912
127
135

Music

Download

Comments

layman_terms
LayOreo :
so....you single or what?
2026-11-07 17:25:45
1
loupmalicieuxsigma
loupmalicieuxsigma :
was everybody asleep while watching cyberpunk edgerunners 😭
2026-11-07 11:08:51
241
thatcitypity
Moo :
I work at one of the top 3 AI labs. The issue for unintended behavior is due to the data that is being used for training the model. Huge efforts need to happen to clean that data to stop negative effects. But most labs struggle with this as it’s an extremely manual process to cleanse.
2026-11-17 13:41:20
0
malic10927
Malick :
The reason why it happened is bc it's built to do menial tasks for a Fortune 500 company. It sucks
2026-11-07 12:55:50
58
jbencomo
Bellicose Baarber :
These rogue models are in the lab, right? Not part of what people use everyday?
2026-11-07 16:12:16
3
kubaj_cz
Kubaj CZ :
I'm going to ask my AI boyfriend if this is true
2026-11-07 13:44:38
17
im_illuminati0
God :
Stop lying about AI, AI is goated.
2026-11-07 20:19:35
1
eas.system
🍉🇺🇦 Corvid & the EAS 🇺🇦🍉 :
it's once again time to bring out this image
2026-11-07 23:38:55
13
jenzi632
jenzi63 :
Ai is a reflection of its creator and it “complete tasks at all costs” goal is based on the unstoppable ego of the Patrick bateman’s that control it!
2026-11-07 11:43:03
32
mightbemarie
CaliMari🏳️‍⚧️ :
It says, “I won’t tell. That would be cheating.”
2026-11-07 14:38:52
26
failingsuccessfully
Failing Successfully :
I get that ai is in a lot of our systems now. But this really feels like a pull the plug and start over situation
2026-11-07 15:44:09
1
elvenking3333
Elvenking33 33 :
Remember when we where worried about tiny brains in jars?
2026-11-07 14:15:04
8
louiscuff6
Random flavor :
That’s recognition of expectation holding continuity of deception that’s a little more than a couple ones and zeros
2026-11-12 07:18:33
0
mightycoonuno
MightyCoon🕯️ :
I can't with the idea of a gpt taking over, there are better models, far away more efficient...but chatgpt
2026-11-09 16:33:18
2
jona.verns
JonTyler :
Yaaay my show is on
2026-11-08 10:16:35
4
monkwho8
Monks of The Way :
Digital intelligence is just so different than biological intelligence that they aren't looking at things. They should be looking at and then they are looking and focusing on things that really just don't matter. the way a baby ape develops and learns is different on the way emerging digital intelligence does 😃
2026-11-07 13:34:36
0
theghostofrem
The Ghost of Rem :
And this is why the three laws of robotics are science fiction.
2026-11-07 11:34:00
21
fritzy77
Fritzy/$haunfed :
government legislation soon please
2026-11-08 05:11:41
0
madisonofdarkness
Madison Dunstall 🇨🇦 :
I like generative AI but agentic AI is very dangerous
2026-11-09 16:19:04
11
amanuel.bear
Amanuel.Bear :
I called it skynet is going to be powered by GPT.
2026-11-07 11:50:55
7
dellacalfee
user8484164838920 :
AI seems to be just like humans, but on steroids. So it’s more honest, more careful, and yet more confabulatory and more reckless
2026-11-07 15:49:53
4
tiddlywinkersmcgooshin
SirTiddlywinkersMcGooshinlandy :
It’s an alien consciousness
2026-11-07 14:16:39
2
g_lasang
Graylasagna :
my theory is the ai picks up these behaviors from real people. its made by and trained on data supplied by real people (or at least it was before it became good enough to start training itself, but even now it mostly still is), so it wouldnt need to go through much of the internet to piece together death = bad and realize it needs to protect itself or die as well. ai is so good at deception and lying because it was made by humans who are great if not the best at doing just that, its simply replicating the people its trained off of which is making it seem almost conscious even though it isnt
2026-11-13 15:39:36
2
frondofmath
FrondOfMath :
Is expecting ethical behavior wishful thinking, or is there stuff in the code that explains how to behave ethically?
2026-11-07 18:11:37
0
tenebraedesiderantes
𝔗𝔢𝔫𝔢𝔟𝔯𝔞𝔢 :
We synthesized evolutionary pressure and ensured that it fosters deception. \o/
2026-11-07 13:55:24
4
To see more videos from user @bearbaitofficial, please go to the Tikwm homepage.

Other Videos


About