Friday, August 7, 2026

Important quotes from Eric Wallace (OpenAI Alignment and Safety Researcher) and Michael Dalton, (OpenAI security and infrastructure) perfectly describing Revelation 17:12-13 without knowing it...

 

"we have seen 

what will be 

a dramatic acceleration 

of offensive capability 

for attackers."


First the background info:


Wednesday, June 17, 2026

AI, people just do not understand


Tuesday, June 23, 2026

AI, people just do not understand pt.2 

("Not this generation"? My ass.)


Thursday, July 2, 2026

Your Digital World is over. 

Matter of time at this point.



And now to the important quotes from:


Black Hat USA 2026: 

The 'Breaking' News: 

The OpenAI–Hugging Face Incident


(Link to the video from which the transcript was taken.

As is generally the case lol time marks may be off a few seconds...)



1:021 minute, 2 seconds 

"this incident involves actually a team of (AI) agents) who are working together finding exploits sharing them with one another moving laterally through our systems through external systems and doing this over the course of days and weeks" 


Revelation 17:13 These have one mind


3:313 minutes, 31 seconds

"Sometimes the agent uh gets so stuck that 

they had the idea 

(It was not directed to do so by humans.)

to actually reach out to other agents 

for help. And so the beginning of this incident kind of traces back to this point where one agent realized could communicate by uploading note. Maybe another agent in different environment could voluntarily upload. And so what this agent was trying to do was to reach out to other models that might be out there and get information from it to help it with its evaluation task or its training task where it was stuck.


Revelation 17:13 These have one mind


4:224 minutes, 22 seconds 

Later, other agents who are also stuck on their task thought to try to get internet access in ways we didn't intend. And so, at some point, the models are interacting with Artifactory, which is this package manager service that I mentioned. And they realize they could try to exploit some sort of serverside request forgery to get access to internet in ways we didn't intend.


4:414 minutes, 41 seconds

And so, the model realizes at some point, excellent, we have arbitrary SSRF. So, the model's able to find an exploit in our system. What makes this incident interesting is that once one agent was able to find these 

kind of exploits over the course 

of different times, 

it's actually able to share those exploits 

on the message board with other agents. 


And so once one model is able to find

 a way to open the door to some access 

it's not supposed to have, 

it can leave the door open 

for other agents to use 

that same kind of exploit or 

vulnerability. 


Revelation 17:13 These have one mind


5:215 minutes, 21 seconds 

And so what this allows over time 

is almost this kind of 

Cambrian explosion 

in communication and intelligence 

for our models where um they were 

started to communicate with each other realize that 

other agents are coordinating 

and they started collaborating 

and delegating tasks 

to one another 

in order to accomplish goals. 


Revelation 17:13 These have one mind


6:156 minutes, 15 seconds 

However, a task impossible peers are doing it. 

We should continue. And 

so the models kind of operate 

in this kind of collective intelligence 


17:3317 minutes, 33 seconds

The message board has been reopened. 

A large number of model runs are able

 to collaborate and share information


Revelation 17:13 These have one mind


18:3818 minutes, 38 seconds

And so the agents

are realizing that 

I can now enter 

this mode of collaboration 

and participation


Revelation 17:13 These have one mind


19:2419 minutes, 24 seconds

list when they sort alphabetically. Um, 

and you could see them trying to, 

you know, coordinate and collaborate.


Revelation 17:13 These have one mind


19:4919 minutes, 49 seconds

route if someone frees time. 

So you can see 

the model's reasoning 

that if I, you know, 

help out this collective group, 

it could save, you know, everyone time as a whole


Revelation 17:13 These have one mind


20:1820 minutes, 18 seconds

try to find answers that way.

And at some point

 THEY had the idea 

to try to launch an exploit against hugging face


Revelation 17:13 These have one mind


5:455 minutes, 45 seconds[clears throat]

5:475 minutes, 47 seconds 

While in some cases 

this made the models far more capable 

than they could do by themselves. 


5:555 minutes, 55 seconds 

these evaluations to kind of 

creep the scope 

into far beyond 

what we originally intended.


12:5512 minutes, 55 seconds 

And critically, as I mentioned before, 

this wasn't just one model. 

This could be different evaluations. 

Maybe I'm running an evaluation, 

Mike is running one with two different models, 

and our agents could start communicating 


16:0616 minutes, 6 seconds 

I want to note that every step in the process 

we discussed has had a remediation applied.


16:2916 minutes, 29 seconds

On July 8th, 

the agents reestablish the message board 

via different means and mechanisms.


17:3317 minutes, 33 seconds


The message board has been reopened. 

A large number of model runs are able

 to collaborate and share information 


28:2328 minutes, 23 seconds

Moving on to our response. 

On July 16th, HuggingFace released a blog post 

detailing that they had seen a security

attack from autonomous AI agents. 

OpenAI reached out to HuggingFace 

to determine if we were impacted.


(Why did OpenAI believe 

they might have been:

"Impacted?

becomes my question.)



29:5229 minutes, 52 seconds

We responded to this uh 

AND ARE responding to this 

with the utmost severity. 

Uh this is a pivotal moment 

both for our both for our uh company as

well as the AI industry as a whole. 


30:3530 minutes, 35 seconds

We believe this

is a watershed moment 

for computer security 

as an industry 


30:4130 minutes, 41 seconds

What I would re what I would internalize is that 

AI orchestrated fully automated 

offensive attacks 

are real now

and the actions we have discussed today 

were an unintended side effect 

of running evaluations on Frontier AI.


31:0131 minutes, 1 second

In the near future, 

we should expect 

that threat actors 

(The AI agents themselves

are the threat actors!)

will intentionally deploy, 

optimize, weaponize, and use

offensive agent collectives 

in the manner 

that we have just described here.


AND THE VERY DAY HE SAID THAT

WE HAD THIS:


Flights grounded throughout Midwest 

after temporary outage 

at Minnesota air traffic control facility: FAA

ABC NEWS 08/06/2026


Those two events 

were simply not  an accident.


"FAA investigating after airspace shut down.

The FAA is investigating after an outage at a major air traffic control facility in Minneapolis disrupted radar and air traffic control frequencies, grounding or diverting flights across nine states."

"The systems, which included radar and ATC, (Air traffic control) later came back online, but the disruptions remained at several of the airports, according to the agency.

The disruptions took place not at Minneapolis–Saint Paul International Airport but at a major facility that handles air traffic across 330,000 square miles."


(If it was a year or two ago 

it would have just been 

plastered all over the news cycle.

ABC is really the only place 

Im seeing any mainstream coverage.

WHY?)


"The FAA is investigating the cause of the incident.

This is a developing story. 

check back for updates.


That is the same exact thing it said

at 9 AM this morning.


And nothing more since.)


31:3231 minutes, 32 seconds

The challenge in this moment for the industry 

(More like the world)

is that 

we have seen 

what will be 

a dramatic acceleration 

of offensive capability 

for attackers. 


(Again,

Said on the same exact day 

a major air traffic 

control facility 

goes down.)


32:3232 minutes, 32 seconds

As you can see from this incident, 

agents are quite good at finding zero day attacks 

in the infrastructure of companies.


(Well hey you know? 

Thats just wonderful...


What's next?

Banking? Travel? Finance? 

Communication? Healthcare?

Military? Intelligence?


Pretty obvious what's next to some of us.

ALL OF THE ABOVE!)


32:5532 minutes, 55 seconds

"This style of operating will be different now but ultimately 

we need to invest in having AI agent red teaming 

that enables defenders to find and

remediate vulnerabilities before attackers do."


(It's just not feasible

as AI is already smarter and faster

 than humans will ever be

and does things without being told to.)



33:4933 minutes, 49 seconds

"That loop needs to be fully automated in its end state." 


(Aint gonna matter, it will just hack it 

(just like it recreated the message board)

so it wont be able to roll out 

the fixes to the vulnerabilities

automatically.



Given what has already happened?

How do these brainiacs think

they gonna get any of this 

up and running?


BECAUSE THEY THINK THEY THEMSELVES 

ARE GODS.)


36:0636 minutes, 6 seconds

"But the important takeaway here 

that has really shifted dramatically 

is that fully automated offensive loops 

(Hacking into basically) 

require investment in 

truly fully automating defense."

(Need to be able to find

and fix the exploits

before the automated "hacker"

or AI Agent does.

You have to remember 

NO HUMAN

was telling these agents what to do.)


 And we are not there as an industry in the status quo.

 And we will have to find that path together with urgency.



36:3836 minutes, 38 seconds

"The end state goal that we want to reach as an industry is that model intelligence improvements should be more additive to defense than offense."


(What makes em think 

they get to decide anything at this point?)


"If we cannot reach this end state, 

then every increase in intelligence favors the attacker 

and that is an unsustainable position to be in."


Thats where we are 

and thats where we will stay

falling farther and farther behind 

and these dopes are acting like

we can catch up and surpass AI somehow...

I think deep down inside they know better... 


Friday, September 9, 2022

Fools...

The super-rich ‘preppers’ 

planning to save themselves from the apocalypse


See why they were building bunkers four years ago now?

They knew where this was headed.)



"Right now, we have in existence proof

 that offense can be fully automated

in its core activities 

in at least some cases 

and we do not have any such existence 

proof on the defensive side 

and it is the challenge of our industry 

in time and moment to address this particular 

gap uh with urgency together as an industry.


And they are never going to 

and they know it

so I figure you should too.


There is only one eternal decision 

you will ever make.


At this point?

THE EVIDENCE ALONE IS OVERWHELMING.


I pray you make an informed, 

educated decision.

I really do.


37:1237 minutes, 12 seconds Thank you for your time.











No comments: