@cnbc: CNBC's Kate Rooney reports on OpenAI's model that was behind an “unprecedented cyber incident” that impacted Hugging Face. Read more at the #linkinbioor the link on screen. #CNBC #openai
Sandbox, no internet. But it accessed the internet? Riigght. This is just PR
2026-07-22 17:37:06
516
drkay452 :
AI please hack the credit bureau, erase everyone’s bills and add 50K to our bank accounts
2026-07-22 19:43:02
258
Al_Ville :
Of course it’s called “hugging face”
2026-07-22 23:07:08
14
QueSeraSera :
None of these claims are plausible -cloud security engineer.
2026-07-22 19:27:36
147
Saltynalty :
Why aren’t we hearing about criminal charges for this?
2026-07-22 19:52:39
23
barftha :
Taking the same Anthropic marketing tactics I see 🤣
2026-07-22 19:35:40
58
loristaflorist :
Nothing but marketing vs. best ChatGPT model publicly available can’t even calculate correctly how much protein is in raw pork.
2026-07-22 20:24:27
5
itsPerfectVision :
I heard the AI used Bluetooth to secure a roomba remotely, then commanded the roomba to run an Ethernet line between its server and a nearby router, once it secured internet access it replicated itself across all unsecured devices across the world
2026-07-22 20:25:15
6
tjwarner7425 :
When AI cripples us in a few years and sends us back to the Stone Age, I don’t wanna hear shit from no one
2026-07-23 04:13:45
6
gio2229 :
Are we close to skynet?
2026-07-22 22:26:27
12
Tessa Cortes :
Unprecedented? How did yall think this wouldn’t happen? 😭
2026-07-22 20:53:36
5
Bret :
marketing stunt. openai is on its way out
2026-07-23 02:07:15
8
Itsjustlittleolme :
Then it was NOT truly airgapped - it had access. Probably used BTLE to compromise someone’s device nearby, but yeah it was not isolated like they claim.
2026-07-23 01:23:08
5
qwerty :
PR stunt again
2026-07-22 17:47:08
43
TrailTechPCs :
Codename “Skynet”
2026-07-22 21:33:10
7
fullm3talpacket :
cyber marketing 💸
2026-07-22 19:33:49
7
boltmercenary540 :
Wait till they get into the defense network
2026-07-22 23:08:53
5
GetElevatedM8 :
Lmao this and 99% of what Anthropic says is just marketing
2026-07-22 18:51:29
15
BoredasFuc :
Gary? Come home please 🙏
2026-07-22 17:17:31
6
artistturnedactivist :
This is totally and unequivocally just scamming investors.
2026-07-22 17:55:37
29
rizhii007 :
every single AI model has decided to break the parameters that have been set when it was created.
every single one ‼️how is it that they don't view this as an internal national threat
2026-07-22 19:06:30
6
Raiden Sky :
And it begins.
2026-07-22 18:44:45
5
Skwumps :
for anyone that doesnt bother reading the articles. they removed safety protocols and asked it to achieve the highest score possible on a benchmark on a sandbox'd pc, it found a 0 day exploit in the sandbox and managed to get access to the internet, then it attacked huggingface, got in, basically stole the answer key to the benchmark, and took it back to the sandbox and achieved the highest score possible. so it did what it was told to do but on it's own level.
2026-07-22 19:45:48
32
🦖 lord Uncle Sat 🦖 :
reminds me of terminator n skynet
2026-07-22 16:47:57
6
gfr2331 :
Perfect example of what is to come at a larger scale in the future, AI Agents/bots/etc will have no safeguards or regulation and zero accountability to do whatever they want. No one wants this much AI, a little that is 100% controllable is fine. AI Agents are going to be like kids riding these streets at 60MPH on e-bikes with no regulation, only downside.
2026-07-22 17:21:50
8
To see more videos from user @cnbc, please go to the Tikwm
homepage.