Subscribe
Share
Share
Embed
Tom, Scott, and Kevin separate speculative AI-doom predictions from the concrete risks created when fast, opaque systems gain authority without credible limits or accountability.
Shared Security is the the longest-running cybersecurity and privacy podcast where industry veterans Tom Eston, Scott Wright, and Kevin Tackett break down the week’s security WTF moments, privacy fails, human mistakes, and “why is this still a problem?” stories — with humor, honesty, and hard-earned real-world experience. Whether you’re a security pro, a privacy advocate, or just here to hear Kevin yell about vendor nonsense, this podcast delivers insights you’ll actually use — and laughs you probably need. Real security talk from people who’ve lived it.
cows will wipe out human race if it happened.
You could have a cow wander into a electrical transmission system,
knock out power for a large portion of the country, of whatever country it is.
You know, as well as I do, that you go 48 hours without power.
People are killing people, right?
It's going to happen.
And that's just going to cycle, right?
Non-zero chance cows will kill humanity.
This week on Shared Security, we pose a question, will AI kill us all?
Well, this is one hell of a question to ask before your morning coffee.
And suddenly everyone, including the mass media and the US government,
is talking about this single question.
Anthropics Dario Amade wants the frontier AI race paced.
And a former anthropologist researcher says we will all be dead
by the end of this decade by a real life skynet from the Terminator movies.
Hasta la vista, baby.
Now, Scott, Kevin and I aren't selling a robot apocalypse countdown here.
We're asking what must be proven before AI gets more autonomy, access and control
and whether anyone, including the AI companies themselves, will actually hit the brakes.
Well, what do you guys think?
Is it time to pack our things and head for the mountains?
Or is this just more AI hype that really doesn't mean anything?
What do you think? Hype or is it real?
Yes and no. Yes and no.
Like right off the bat, I believe and have long stood by the statement
that saying something has a non-zero chance of happening is a less measurement.
It is as useful as me saying, hey, Tom,
you have a greater than average number of legs, right?
Which because the average legs on humans is less than two.
All right.
And I know other people sitting there listening right now going, no,
almost everybody has two legs.
I'm like, I understand that.
But the way averages work, that's right.
Right. Unless anybody has the exact same number,
the number, the average number will be lower than the biggest number.
Right. Just math.
So saying we have a non-zero.
OK. Yes.
You know what else we have a non-zero chance
of killing the entire human race cows.
There is a non-zero chain.
OK. That cows will wipe out human race.
If it happened, you could have a cow wander into a
electrical transmission system, knock out power for a large portion
of the country of whatever country it is.
You know, as well as I do, we've all seen the movie Purge
that you go 48 hours without power.
People are killing people, right?
I just it's going to happen and that's just going to cycle, right?
Non-zero chance cows will kill humanity.
So I just end with all the things happening with, you know,
going on around now, as as Brittany just messaged me,
because he used meat.
I'm surprised we haven't had to be cold yet of that.
Do we know that's not caused by some rogue AI that's.
I'll tell you right now, it was because rogue AI
and that rogue AI is named RFK Junior.
Oh, sorry.
I set you up for that one, Kevin.
You did. I appreciate that.
I know that here's here's my opinion.
I think that there's two problems with this introduction.
And I don't mean your introduction.
I think your introduction was awesome.
One, the idea of non-zero chance.
Of course, there's a non-zero chance.
We have systems that people are using to make decisions,
even if we don't give the AI system the autonomy to do something, right?
How many people here have seen the episode of the office
where Michael and Dwight drive into the lake
because the GPS told them to turn there?
Right. Yeah. Yeah.
You can't convince me that Hagswith or whoever
isn't asking Claude something and having Claude say,
you'll be wall and they would do it, right?
Of course. Yeah.
OK, so non-zero chance.
Here's the other problem I have with this entire thing.
And we keep falling for this stuff, right?
We've talked about this before.
I think it was Google we talked about before.
Every single time knows Facebook, every single time
an organization is closer to a monopoly or success
in their field and they're being chased
by a whole bunch of smaller competitors.
All of a sudden, they demand legislation
against the thing they're doing. Yeah, it's so crazy.
Yeah. The only organization
that is going to be helped by legislating and limiting
our research into AI is Anthropic and Open AI.
Period. Right.
And I'm not saying we shouldn't have guardrails.
I'm not saying we shouldn't have something.
We all know my opinion of legislation.
But every single time you hear an organization that says
because here's what they they say part of this
and the other part is just kind of implied.
Hey, we've spent a lot of money building something.
And because we've built it,
we've proved it's possible.
So other people are now trying to build it.
And sadly for us, it will be easier
for them to build it because we've proven the path.
So what we want to do is spike the path
so they can't come in. Yeah. Right.
And the way they spike the path is by going to Congress
and saying, hey, we paid you a bunch of money.
I mean, we didn't pay you.
We donated to your election campaign
past this law that makes it harder for everybody else.
Yeah. I'm tired of that.
And I'm tired of the fact that we as a population
keep just accepting that call.
The fact that people I know that have.
They could not spell AI are now saying,
well, we should have laws protecting us from this AI system.
I look a cool.
And how many laws have protected us against murder?
How many law rights? Sorry.
I'm right. Yeah. No, no, it's a it's a good rant.
And I agree with you, Kevin.
I think I find it funny
that the value of these companies
just continue to increase every time
a founder like Dario, whoever comes out and says,
AI needs to be paused, needs to be regulated, needs to.
We need to stop what we're doing.
We have to do something about it and put guardrails in.
And like, I'm like, well, first of all,
the investors just keep throwing more money
and billions of dollars at these companies.
The more they say that.
And then second, they're the ones creating these things,
but they're not pausing and they're not stopping.
They're continuing to do this.
So to me, it's all marketing and it's about money and investment.
Absolutely.
The fact that they're getting more investment
when they're literally saying government make us stop
tells me that the investors know they're not going to
and the law is not going to affect them.
I have this great idea if Dario or whatever.
And for some reason, I just see him as a Italian plumber in a game.
Every time I hear his name, Mario, Mario and Mario Dario,
Evil twin brother, Dario and every time I see these,
like, I want to write back to the dude and say,
I got this great idea.
I got this great idea, man.
You believe that I should be regulated.
So give me a list of the things you want the government
to make it illegal for AI to do.
OK, cool. Give me that list.
Now, tell me why you haven't stopped your own damn company.
You know, the one you're in charge of,
you know, the one you make the rules for,
the one that you claim is just trying to help population
and humanity will implement those rules, jerk, right?
The day you show me that you have taken your proposed
regulation and made it not policy,
I want to see actual controls that prevent the thing
you're claiming and everybody else is going to do.
Show me those controls.
And the day you show me those controls and show me
that they're effective, then I will have an actual
conversation with you about regulation.
But until you show me that you're you're putting
your money where your mouth is, because, like I said,
you control the organization.
That would be like me saying, you know what?
Secure ideas has this really serious problem
that our consultants are becoming artificially intelligent
and they may destroy the world.
But if the government doesn't regulate me from
preventing them from destroying the world, look,
my consultants are going to go crazy.
Yeah. I'm sorry.
We told you. My hands are tied.
I got to be destructive.
Only the government can save you from secure ideas.
It's BS. I'm sorry. It's BS.
OK, I'll shut up now.
Scott, what do you think of all this?
Well, I mean, I can't create as entertaining
an argument as Kevin on whether or not
things should be regulated.
So I'm just going to leave it at that.
But I think there are a lot of different paths
by which I could severely harm the human race.
And I'm just going to go down that path.
Yeah, there you go.
And say, you know, the recent,
you know, evolutions of, you know,
breaking out of testing sandboxes
to me is the most credible.
A lot of people will say, yeah, it's it's marketing.
You know, they did it to make their
things seem more mysterious and powerful.
But I can think of other ways that they might have come up with,
you know, a killer app that would grow their market,
maybe not as fast or I don't know.
I'm just saying I think that the most credible way
that this stuff gets to that point of being,
you know, a serious threat to all humanity is the way it's going.
In the path that we've seen it going on.
And I think, you know, some of the scariest things we could see,
if my feeling is correct, is it would be finding paths
to a distributed way of operating
so you don't have a kill switch, right?
That would it would looking back to Asimov's, you know,
three rules of robotics, right?
You know, did you see the XKCD cartoon about that?
You got to find it.
Why Asimov put his rules in the order that he did.
It's actually the picture of the week
and last week's security now episode from Steve Gibson.
But basically every other combination of don't harm humans,
obey the rules and then, you know, protect yourself, right?
Every other combination results in bad things.
I don't know how I'm not an expert on this stuff.
But I think the process starts with identifying, you know,
is this going in a direction that could, you know,
lead to what they're predicting?
And I think it may be.
And I've always sort of had that gut feeling that, you know,
it was going to start with, well, you're not going to be able
to trust what you see on TV or on, you know, streaming.
And that's happening.
It's happening now.
Yeah. Yeah. Yeah.
So, yeah, that's that's my two cents.
And I want to be very, Scott, agree 100 percent with you.
I'm not saying the uselessness of the measure
means we should ignore the fact that it's possible.
It is absolutely possible.
As a matter of fact, I would argue possible isn't the right word.
I think the word should be likely, right?
Whether it's directly due to the AI system going like,
hey, wait, these humans suck, let's wipe them out.
Or it's just because we've moved to something
that either people are making decisions based
on what AI systems are doing or we're allowing AI systems
to control things and making judgment calls
based on incorrect information, but we're making them at speed.
Right. Because that's that's the thing.
Look, I'm dull, but I can only be dumb so fast.
Right. AI could be dumb just like me way faster.
And if I take my level of dumb and combine it with the speed
of AI giving me more dumb, we have an issue.
So yeah, I totally think it is possible
and likely that we, even if we don't wipe out humanity, right?
The idea that we cause significant harm
directly due to the usage of AI, not
data centers sucking up all the waters and stuff like that.
But directly from the usage of AI, I took AI
and hosted up to this nuclear power plant and it went right in the meltdown.
Or it's even less obvious than that.
I think, you know, that method I said of getting to a distributed operation.
What about what about the botnets that have taken over all the old routers?
Right. Hmm.
Isn't that a logical way that they might go, right?
Rather than first attacking a nuclear reactor?
Well, I'm not saying attacking a nuclear reactor.
I'm saying we give it control to the nuclear reactor and it makes a mistake.
Yeah. Oh, I see. Yeah. Yeah.
Yeah. Accidentally, I guess I'm more of a realist
that like we don't need AI to destroy humanity.
Like we're already doing pretty well.
It's like we're like, yeah, like literally like we got climate change.
We have, you know, I mean, pick your poison, right?
Like I'm actually more fearful of, you know, World War Three
and all of us launching nukes at each other.
And there definitely is the end of humanity and AI won't even be involved in it.
Right. And I see this all over the media and the news.
First of all, the odds of this happening.
Yes, there's always a chance and all these things.
Well, it's going to happen in five, 10 years.
Like who really knows?
Like most likely not.
I mean, and I think it is kind of overhyped,
but I think we need to direct our attention towards the things that we can control.
Right. Like and realize that this AI hype is really all of a lot of it
is built around money and investments.
And we probably don't have time to talk about the whole US versus China thing.
And Trump saying is like, well, we can't have guardrails.
We got to beat China.
Guys, China's already won.
Really sharp president.
No, seriously, China's already ahead of us.
Like we should just admit that.
But we need to focus back on what we can control.
And it may not be AI turning into the next Skynet.
Just my opinion.
Yeah, priorities.
This one is a hotly a hot topic.
And we all have very, very strong opinions about it.
But love to hear from our audience.
What do you think?
Do you think this is the end of the world as we know it?
And all of humanity will be wiped out in five years?
Or do you think it's a more hype and more marketing and more about money?
And you put the REM song in the background.
I could maybe on the audio version.
But YouTube will ban me for life if I use the copyright.
So hopefully me seeing it just now won't get you banned.
But, you know, yeah, well, I may have to edit that out.
Scott, thank you.
Do you have any questions to anyone?
Yes, yes, that's right.
Please let us know in the comments.
We've been getting some great comments on the on the YouTube channel recently.
Even on past posts, I saw a really good one about remember the topic
we had on AI boyfriends and girlfriends.
And someone left a very good, well thought out comments about that.
And I was good about, like, we should be really focused on human
relationships with each other and not so much relationships
with AI systems that, well, will never be human.
There's some good studies on that.
And maybe we may have to discuss again.
Yeah, that's right.
That's right.
Yes.
All right.
Well, we will move on from this hot topic.
But again, thank you all for listening.
And until next time, stay safe, stay secure and stay private.
Thank you for listening or watching.
If you like this episode, hit subscribe, share it with your friends and
colleagues or jump into our community at sharedsecurity.net
slash supporter to keep the conversation going.
Thanks again, and we'll see you next week for another episode of Shared Security.