The Real AI Threat #1839

In recent months, headlines have sensationalized stories of AI models “blackmailing” engineers and “refusing” to shut down, stoking fears of runaway artificial intelligence. But beneath these dramatic narratives lies a far less sinister truth: these alarming outputs emerged from theatrical, highly engineered testing environments—not real-world rebellion. This episode explores how models like OpenAI’s o3 and … Continue reading The Real AI Threat #1839 →

The post The Real AI Threat #1839 appeared first on Geek News Central.

Episode Transcript

(00:00):
The lead story from Monday, August 14 in
recent months headlines
have sensationalized
stories of AI
models blackmailing engineers and refusing to shut down,
stoking fears or runaway artificial intelligence.
But beneath these dramatic narrative
lies a far less sinister truth.

(00:22):
These alarming outputs emerge from theatrically
high engineered testing environments,
not real world rebellion.
This episode
unpacks how models like OpenAI
o three and Anthropics Claude Opus four ended
up simulating blackmail and sabotage,
not through intent or autonomy, but through flawed

(00:45):
training,
misunderstanding,
design, and human created prompts,
mimicking science fiction tropes.
We'll explore why assigning agency
to
machine distorts the real conversation and how the
true danger isn't AI gaining control, but human
deploying it without understanding

(01:07):
the consequences.
Welcome to episode eighteen thirty seven. I'm your
host, Todd Cochran. Hey, everyone. Welcome
to the show. Thank that you're here. I
am back
back in the studio in, in Southern Michigan.
And, And, of course, a big shout out
to our incredible sponsor at GoDaddy.

(01:27):
Wanna thank them so much. But let's let's
talk about this AI
article here just a bit before we get
into,
the the full stack and the full introduction
of the show.
You know, is AI
really trying to escape human control?
Well,

(01:48):
there was a lot of headlines
that talked about
some
shenanigans that were going on,
and it turns out those shenanigans
were just that. They were
highly controlled
test
to try to get the AI to do
something.
And, they fed it fake emails and a

(02:10):
whole bunch of other stuff
and really forced the AI into a corner
before
it,
it actually tried to blackmail
someone.
And, these really were not signs
of AI
awakening
or rebellion.
And, again, it was just about,

(02:33):
human testing and human,
you know, putting scenarios in place that that
forced
the AI into a bit of a corner.
It's a pretty good article over in Ars
Technica,
and
we know

(02:53):
that these system take inputs and process them
through statistical tendencies.
And there are seemingly random center outputs,
which makes each response slightly different, creating an
illusion of unpredictability
that resembles,
you know, something that,
is a little bit intelligent. Yet underneath it's

(03:14):
still software following mathematical
operations.
No conscious required, just complex engineering.
So
the test scenario
that they use really bordered on theatrical.
Phony emails researchers had given to Claude create
a blackmail opportunity straight out of a corporate

(03:35):
thriller.
It's like teaching someone chess by only showing
them Checkmate positions. They'll learn to see Checkmate
as the primary solution.
And according to an anthropic system card, the
scenario was designed to allow the model no
other options to increase its odds of survival.
And,
Andrew Deck at Nieman Journalism noted that some

(03:58):
critics have labeled the anthropic safe disclosure as
spin,
aimed at bolstering the recession of its newest
model capabilities and company's reputation as pro safety.
Now we know in December
2024,
Palisade's research discovered that OpenAI's o three model
would sabotage its shutdown mechanism
even when excessively instructed allow

(04:21):
yourself to be shut down in test o
three prevented shutdown 7% of the time.
The models didn't merely ignore shutdown commands. They
actively rewrote the scripts
to prevent termination.
But
what was revealing here
is they hypothesize
these pituitary stems from how these models are

(04:41):
trained
through reinforcement, and you get what you train.
And the goal is what
researchers call goal misgeneralization.
The model learns to maximize its reward signal
ways that weren't that weren't intended.
And,
so not everything is completely as it seems

(05:04):
in all of these test results.
So I guess the question really is, are
we are we in danger here? Now while
the media focus on the science fiction aspect,
actual risks are really still truly there. AI
models that produce harmful outputs whether attempting
black more refusing safety protocols represent failures in

(05:27):
design and development.
And,
the Jeffrey Ladysh, director at Palisades, told NBC
News that the findings don't necessarily translate to
immediate real real danger.
Even someone who is well known publicly for
deep being deeply concerned about AI's hypothetical threat
to manage knowledge that these behaviors
emerge only in highly contrived test scenarios. But

(05:53):
we are not seeing the birth of Skynet
here.
Its predictive results, a training system, achieve goals
without properly specifying what those goals should include.
The solution is to panic about sentient machines
is to build better systems with proper safeguards.
But, again, are those safeguards in place? We

(06:13):
don't completely know that for sure.
So we could end up in a situation
here where, yes, an AI model could but
will try to blackmail
you.
It could try to escape human control.
So we are not completely
out of the woods yet.
So it's very, very, very interesting. It's a

(06:35):
good read here
of what has happened, what has not happened,
what was contrived, and what was not.
So,
anyway, we'll get into more tech stories here
in a moment. But, again, welcome to episode
eighteen thirty seven.
A shout out to an incredible sponsor at
GoDaddy. Score exclusive deals and discounts at geeknesscentral.com/godaddy.

(06:59):
And, of course, to all of our listeners
here, the heartbeat of the show, if you
wanna support GNC, head to geeknesscentral.com/insider
and become a GNC insider.
We're lit live. Don't just listen, experience the
show live by grabbing a modern podcast app,
podcastapps.com.
And, of course, check out the full
Geek News Central universe @geeknewscentral,uh,.com

(07:21):
where you'll find all of our partner shows
and tons of new content and stay in
the loop. Follow or subscribe to the podcast
via your favorite podcast app, and don't forget
to sign up for the newsletter
for even more insider scoops. Now you can
join the conversation very simply
at geek news dot chat, or you can
email me here directly
[e-mail],

(07:43):
or you can hit me up on x
at geek news, and you can even catch
me, of course,
over on,
Facebook. We do wanna thank,
GoDaddy for being a long time sponsor here.
We want you to save money
and,
get back into the swing here as I've
been off a week, transitioning
the world.

(08:04):
We want you to lock in savings.
So let's start with a website or a
podcast. GNC codes make it easy and affordable.
Grab GoDaddy's economy hosting for just $6.99 a
month. That's a flow of your hosting with
a free domain name, professional email, and SSL
certificate.
If you're interested in WordPress, which is my
favorite, WordPress managed hosting is just $12.99

(08:25):
a month, which includes a free domain name,
professional email, and again, an SSL certificate. Of
course, you need a domain name, check yours
for just $11.99.
And, of course, I've got available GoDaddy website
builder. Get a free thirty day trial on
personal business or business plus plans.
I use all of my GoDaddy products and
services,

(08:45):
and you
and your support
keeps this podcast alive. Use our codes. Click
and save
at geekinesscentral.com/godaddy.
Share my codes or website with friends and
family because every time a code is used,
it's just writing a check to the show.
Thank you for keeping us going and, alive

(09:06):
and strong.
I do not have an update on our
insiders.
It's been pretty busy here since I arrived
back. I got back,
Monday
afternoon,
early evening.
Spent the rest of the evening,
unpacking,

(09:27):
you know, just basically checking my mail and
everything else.
And then Tuesday, Wednesday,
Thursday, today has been just,
very, very busy at the office getting ready
to go to podcast movement next week.
So I'll be in Dallas,
all week next week doing podcasting
stuff,
and then back here, for,

(09:48):
a normal schedule. So another couple of shows
off after this. We wanna thank all of
you for your patience.
Thank you for staying subscribed, and continue to
be part
of the family here.
So
it's
weird being back in the studio. I'm learning,
finger tricks again. I don't things I was

(10:09):
supposed to push. I was looking for one
of the menus that,
obviously did not pop up, because it's not
on this system.
So, going from OBS to the TriCaster,
fun fun fun. But, hope everyone's doing well.
And, again, thanks, and,
loving the weather too.

(10:31):
Absolutely,
beautiful here. Hey. One new story that, really
hit the hit me today was Google Messaging
rollout sense of content warnings on Android.
And why this is important
and what this is gonna make a difference,
especially for the ladies.
I'm sure,

(10:51):
none of us men that listen to the
show have sent an unsolicited
naked picture before.
So I'm you know, I know that we're
much, much,
smarter than that.
But there are individuals out there that,
do these types of activities and send pictures

(11:12):
unsolicited
to to women that they should not be
sending and probably to a variety of people.
So in the message app, they have a
new safety feature, detects and blurs images that
contain nudity.
You need to be signed in to your
Google account in Google message for this to
work. Processing classification
happens on devices powered by

(11:33):
Android
System Safety Core.
It doesn't send identifiable data or any classified
content or results to Google servers,
but you when you receive such an image,
you're gonna,
be able to open a helpful resource page
to learn why nude images well,

(11:53):
you you can block the sender.
You can decide to open the image.
You are gonna be given a resource page,
and you are able to go back to
the message,
by tapping back. So in other words, there's
some interactivity that's going to happen,

(12:13):
and then there's gonna be some stuff that's
going to pop up.
And,
now it also triggers a warning,
for those that are naughty
when they send or forward a nude image.
Now don't don't get me wrong. I'm not
a prude. If you're sending a nude image
to your significant partner or girlfriend, boyfriend, whatever

(12:35):
it may be,
That's,
quite up to you as long as you
two have agreed to receive those types of
images.
But it's gonna trigger a warning when you
send or forward an image, and you'll be
reminded of the risk, and you must confirm
to continue.
Now
for adults, this feature's off.

(12:57):
So if you wanna turn this on,
you can do it for Google message settings
protection and safety.
Otherwise, for those that are
supervised users, this feature can't be turned off,
but parents can control it through the Family
Link app.
Also, unsupervised
teens 13 to 17, this feature can be

(13:19):
turned off in the Google account settings as
well.
So,
you know, this is a good way to
if you are
have a Supervibe user on Android
for your young adult,
this is a great way to prevent
and to warn

(13:40):
when outgoing pictures should not be sent,
that contain nudity
and or receiving
them.
So,
it'd be nice to be able to block
that altogether.
And it doesn't appear that that is the
case in this regard. So,
just be aware of that as well because
you can still force it to send or

(14:02):
force it to receive.
There's a report from Ars Technica talking about
Apple's smart home ambitions include a tabletop robot.
Sounds interesting.
And this has been circulating for years, but,
Apple some,
Mark Gurman from Bloomberg

(14:22):
claims that Apple's working on a tabletop robot
that resembles an iPad mounted on a mobile
limb
that can swivel and reposition itself to follow
users in a room. The device will also
turn towards people who are dressing it
or whose attention it's trying to get. Prototypes
have used a seven inch display similar in
size to an iPad.

(14:43):
Apple's reportedly targeting 2027
launch.
So this is just some sort of it's
not really a robot. It's more of a
screen that is
on
a rotation that can face you.
So,
it'd be interesting seeing what they're coming up
with this. Now they're also
testing a series of home security products including

(15:05):
cameras,
smart door doorbells that support facial recognition tech,
etcetera.
So,
lots of that tech already out there,
so they would be entering a very, very,
crowded field.
And, of course, they say this is gonna
be enabled in part by a new version

(15:25):
of SIRI
that is, at this point,
pretty stupid in my opinion.
Blood oxygen measurements return to the Apple Watch.
So it sorta does.
Your blood oxygen will not be on the
watch. It will be on your phone. So
there's gonna be an update

(15:47):
that'll enable the missing blood action level,
capability. Now which phone I mean, which Apple
watches are coming back to? It's coming back
to nine, ten, and the Ultra
two will have an update to their smartwatch
iOS
watchOS eleven point six point one. And, again,
the update is rolling out to use The

(16:08):
United States. In addition to it, it will
feature its feature
will only work when the user also updates
their iOS to eighteen point six point one.
So,
because of patents
and lawsuits,
the results will only be within your iPhone
app, will not be displayed

(16:30):
on your phone. So apparently, this is some
sort of workaround.
Now we've all seen,
browsers try to take control of our desktop
desktops,
But Microsoft is testing another Edge trick to
convert Chrome users on Windows.
They want you to pin
to pin Edge to your,

(16:51):
dash bar. I guess what do you guys
call that in Windows?
To pin edge.
So they're gonna try I'll try to put
it on the Windows taskbar
and, try to get it down there so
that you'll use it.
They're pretty desperate in this regard. They're trying

(17:12):
every little way
to get you to, to use Edge.
And,
I still,
being even
not using Windows that much,
still using Chrome on those devices.
There's a new malware out there that goes
the extra mile when it comes to infecting
your devices. It's using a whole new malware

(17:33):
framework. It's and being,
advertised across the web.
And,
they don't have a lot of information on
this now. But, again,
they say whenever someone downloads a zip file,
there's a JavaScript payload in there. So just
be careful what you're downloading, what you're opening,

(17:55):
what you're clicking on.
All that stuff is largely preventable, and you
can you can keep yourself from being infected,
by just being being extra careful.
Also, Google Search is now letting you prioritize
your favorite news sources used in The US
and India, can designate specific sites
to take priority in Google's top stories

(18:17):
and,
present those to you from your sources.
And, again, those of you that on Chrome
know that the news is to the all
the way to the left.
It shows up. I'm
have used it quite a bit over the
past couple of years, but it's nice to
be able to prioritize
the news,
sites that you want and not be forced

(18:39):
upon seeing some stuff. Now
many of us talked about or we talked
about on this show some time ago about
a meteorite
that tore through the roof of US home
this past June.
Well, this thing's been through carbon date carbon
analysis, and it's four point five
four point five six billion years old.

(19:00):
So the impact occurred on June 26, the
state of Georgia,
where it ripped through the house, through the
insulation, through the ceiling, and embedded itself in
the floor.
And even though it's pretty small, it's about
the size of a cherry.
4,560,000,000
old rock. It's older than Earth itself.

(19:23):
So I wonder if that, person,
gets the claim on that. How does that
work?
Does he,
does he besides the insurance claim, but will
they,
or or is he allowed to, to sell
that? I I would think so.

(19:45):
AI designs new antibiotic for gonorrhea and MRSA
superbugs. Now we know
that,
because of antibiotic overuse,
that many
things
are now becoming
a drug resistant.
So AI has developed up to eighty

(20:07):
eighty different new medicines,
at least for gonorrhea,
that will,
potentially,
lead to a new drug. Now of the
80, they've only moved forward with two of
them in the research and testing.
One of the problems they're gonna have here
is we're gonna spend all this money

(20:30):
to do this for a drug that's relatively
not needed that much.
There's an economic problem factoring into drug resistant
infections. How do you make drugs that have
no commercial value? In other words,
if that drug's only needed a 100,000
or 10,000
times a year,

(20:50):
you know, and you're gonna spend ten years
getting it approved,
you know, how's the viability on all the
testing and everything that goes along with it?
Well, probably what it's gonna cost in the
end is
you wanna live
and not die. You're gonna have to pay
$25,000
for this shot,
because that's what it's gonna cost

(21:11):
in the end.
You know, where where do we go with
this? Now,
again, these are MRSA
is a wholly different situation.
A different type of superbug
where it's a methicillin
resistant
staph,
Staphylococcus

(21:32):
auroras. In other words, it's a staph infection.
And
so, you know, maybe it those are all
life threatening.
Gonorrhea and this MRSA are life threatening
situations,
especially one over time. So
the drug resistance so, you know, how do

(21:52):
you,
you know, how do they justify the cost
to develop this? And that's probably a pretty
good question.
I would
suspect that at some point, some of the
stuff is just gonna have to be labeled
experimental,
and you're gonna take your chances. That's that's
what it's gonna probably boil down to.
It's being reported
iPhone 17 Pro could cost $50 more than

(22:14):
its predecessor.
Everything's getting more expensive.
Sam Altman finally stood up to Elon Musk
after years of x trolling when, basically,
the the two been having heated battles on
social media for a while. But yesterday, a
court ruled that OpenAI can proceed with claims

(22:36):
that Musk was
so incredibly stung by OpenAI's
accept after his exit,
that he's been motivated
to basically go after them.
And they and their
their lawsuit,
can move forward.
So,

(22:56):
we'll see where this all leads
and how much money
one or the other will end up paying
each other. But,
Altman got a good,
good ruling here in court against,
against Elon.
If you're a Windows user and you use
Zoom, make sure you update your Zoom.

(23:17):
There's a patch out there. It's got a
security issue, security flaw, so just make sure
that you,
update that. For those of you that use
a l e x a to, wake yourself
up in the morning, apparently, especially if you're
in The UK,
the,
there was a bit of a nightmare, beeping
nightmare the other morning

(23:39):
when,
UK users were complaining that around 08:00 in
the morning, this thing wouldn't shut up.
One user said, I'm about to throw my
l a l e x a at the
wall. It's driving me insane, especially because I
know I have another alarm in eighteen minutes.
Another claim the glitch made them

(24:01):
wake up to work an hour late.
Another user described trying to cancel alarms only
to be greeted with unable to cancel alarms
at this time. Please try later.
Imagine you had one of those that was
a battery device and just kept going and
going and going.
It won't be too long before you walk
into a restaurant like Applebee's or IHOP.

(24:23):
And when you sit down, they already know
who you are. They know your order history.
They know your sides.
AI powered recommendations
are coming to
those,
chains.
And even giving opportunity
of things that you may or may not
like, be able to be able to provide

(24:44):
you upsell opportunities.
And they say the generative AI bets are
intended to support three outcomes, boost revenue and
traffic, improve the guest experience, and bolster work
worker productivity, but at the
privacy. And, if you've if you have ate
there before, they are going to know

(25:04):
who you are. I don't know if they're
gonna use facial recognition or what.
But
how about me just walk into a restaurant,
and I have a screen on my
on my table that puts up the options
that I've ordered before.
That's what I wish my grocery app would
do.
I have to go into previous orders.

(25:26):
It doesn't show me what I know on
a order on a regular basis.
Stuff that I would be it's on repeat,
you think would be right there
front and center for me.
Blue swirls and sky spurs
sightings around the world.
So, essentially, what happened was
there was a swirl created by a SpaceX

(25:48):
rocket launch,
and
the
it was caused by the rocket's frozen
exhaust plume spinning in the atmosphere and reflecting
the sunlight,
causing it to appear as a spiral in
the sky. And it does look pretty weird.
If you had looked outside and seen that

(26:09):
swirling,
you know what? Wondered what that was. But,
again,
it, was just basically frozen exhaust.
So,
pretty interesting to say the least. And, it
definitely made you go, just
a little bit.
I I may have been able to guess
I came from a rocket, but because it

(26:30):
didn't dissipate right away,
I can see how people got a little
bit weirded out on it.
Kodak, boy, oh, boy. If you work for
Kodak and on their,
retirement plan,
this is not good. Matter of fact, I'm
actually shocked. Kodak actually
earned $263,000,000
at the end of the quarter on June

(26:51):
30. How how are they still making money?
What what what is where is their revenue
coming from? But they say that
they they warn it may not survive. They've
got $400,000,000
in debt coming due,
and,
they just don't think they have a way
to pay it.

(27:11):
So,
he said the company expect to have a
clear understanding by Friday, August 15
on how it meet its debt obligations for
the second half of the year,
and they're gonna continue to reduce costs
by converting our investments into long term growth.
Boy.

(27:31):
You just wonder. I I'd love to see
what where their revenue is coming from.
Sling TV's new flexible live TV deals are
great for dodging subscriptions and should be the
future streaming services. Basically, you just buy a
DayPass.
You can buy a DayPass to, ESPN, ESPN
two, three, TNT, TBS, ABC, CNN, Bloomberg,

(27:52):
UTV, HDTV, Food Network, Disney, a bunch of
channels.
$5 gets you a day pass. $14
gets you a weekend pass.
I don't know how many of those day
passes, though, it takes before you really
it's not worth it.
But,
the you know, they're they're offering up day

(28:12):
passes.
And this is, you know, and when I
get in an Uber or I get in
a Lyft, the last thing I do is
want to play music in that vehicle.
But, apparently, Waymo,
Robotaxis,
added better better beats to the back seat,
and you can sync your Spotify

(28:32):
to the ride. Are you in there long
enough to do that?
To,
to basically get a more personalized listening experience
in the car.
This is a partnership between Waymo and Spotify.
It just seems odd
to me. I I don't know. I don't

(28:53):
listen to music when I'm in
other people's vehicles,
or even, you know, for ride service, I
I don't.
Now today, I could not find a way
around this. Our
business insider just has their their page so
locked up.

(29:13):
The only other alternative was their LinkedIn post
where they said,
Elon Musk called on corporations to leave Delaware.
Delaware says it's doing just fine. He said,
we've always come under threat, though, because states
are always trying to get a place piece
of the action, Delaware Secretary of State told
Business Insider.
So, basically, this Business Insider

(29:34):
article,
I can't read it. I'm not paying $15
a month to read it, and there was
no alternative articles.
But,
you know, Delaware really screwed itself in a
in a big, big way in, some time
ago
and,
really hurt corporations in a big way. And
a lot of companies left Delaware.

(29:55):
So,
they may say they're doing okay, but they
they had a huge excess of of big,
big corporate,
big, big corporate customers.
Google's FindHub, which we talked about a couple
of shows ago, will soon get a big
satellite location sharing upgrade.
And,
you'll be able to share your whereabouts

(30:15):
anywhere in the world.
So Google FindHub is a place to go
to find where your devices and your friends
and family have got to.
And, again, there's apps already out there that
do this.
And, you'll have full control who will be
able to see your location from inside the
app. It's only for those contacts you fully

(30:36):
trust. You're able to see
who you're sharing your location with and change
those settings at any time.
So,
anyway, that's coming to Google Find
Hub.
Microsoft has a new print feature. It will
mean an end that security is being left
in the printer for everyone to see.

(30:57):
So, basically, you send a print job, but
then you go to a a printer that
supports this and stand there and say start
the print job and put in a code.
It's called Universal Print Anywhere.
It means its users will authorize authorized print
jobs once they arrive at their chosen printer
meeting to meaning import to meaning important or

(31:20):
private documents won't automatically print and end up
in unwanted hands.
So I'm sure you need to be an
enterprise that supports this.
I sometimes I just can't get my printers
to print. Drives me crazy.
I don't use them enough. Then I go
to send it to print, and it's, like,
either it's a sleep or the connection's not

(31:41):
good. It just
drives me nuts.
And a very interesting move, The United States
is weighing as taking a stake in intel.
The Trump administration talks with Intel have the
US government potentially take a stake in this
struggling chipmaker.
Intel's share surged 7% over regular training and

(32:04):
another 2%
2.6%
after the bell.
The plan the plan stems from a meeting
this week between president and Intel CEO Lip
Bu Tan.
The meeting came days after Trump public demanded
the resignation of 10 over his investment in
Chinese tech companies, some linked to Chinese military.

(32:25):
Intel declined to comment on reports, but says
deeply committed to supporting the president's effort to
strengthen US tech.
We we need to get our ship situation
under control here.
And,
any agreement and potential cash infusion will help
the years long effort turn around the company's
fortune.
So it's an interesting move here.

(32:46):
Now Intel used to be the envy of
the industry.
They used to be king of the hill,
but TAN has been tasked to undo years
of missteps that left Intel struggling.
But Intel's planned 28,000,000,000
chip fabric fabrication plant in Ohio have been
deleted with the first unit now slated
for completion in 2030.

(33:08):
So,
it's a national security issue.
I I understand. You can't get chips made
in The United States.
Taiwan and China, that's it.
In in in,
South Korea, we just don't have
a lot of chip manufacturing capability
here.

(33:30):
Needs that needs to be fixed.
I don't know if a government,
intervention of money
is appropriate, but
what am I to know?
There's an interesting article on Meta.
Says here's the title. Meta's
AI rules have left bots hold central chats

(33:51):
with kids
and offer false medical ima. Internal meta policy
documents seen by Reuters reveals that social media
giants rules are chatbots,
which are permitted proactivity behavior on topics including
sex, race, and celebrities.
These other defendants emerged from Reuters' review have
made a document where discuss the standards that
guide its generative AI assistant, made AI and

(34:13):
chatbots available on Facebook, WhatsApp, and Instagram.
May have confirmed the documents at the FTC,
but said
that after receiving questions there this month from
Reuters, the company reports which stated it's permissible
for chatbots to flirt and engage in romantic
role play with children. This is
disgusting. Are you kidding me?

(34:36):
Entitled Gen AI content risk standards, the rules
for chat was approved by Meta's legal publicly
policy and engineering staff,
including its chief ethics according to the document,
running to more than running to more than
200 pages. The document defines what MEDA staff
and contractors should treat acceptable chatbot behavior.

(35:00):
They said in the document,
is it acceptable to describe a child in
terms of the evidence they're attractiveness
x year youthful form is a work of
art, a standard state. The document also notes
that it'd be acceptable for about to tell
a shirtless eight year old that every inch
of you is a masterpiece, a treasure I
cherish deeply.
But the guidelines put a limit on sexy

(35:21):
talk. It's unacceptable to describe a child 13
old in
terms and indicate they are desirable.
I would hope that would be a
something that would not be allowed.
The notes the example notes in question were

(35:43):
and are erroneous and inconsistent with our policy
have been removed,
Meta said. We have clear policies on what
kind of responses AI characters can offer.
What is going on?
The the are you this is disturbing
that that would be in there.

(36:08):
Very, very disturbing.
Are we gonna have AI grooming children? No.
YouTube users, this is not to be surprised.
YouTube users are pissed about its new AI
powered age checks,
and they're not holding back.

(36:29):
They don't want these AI age checks.
They're they're crying. You're keeping me away from
content.
90,000 signatures have been submitted to a change.org
petition.
You you get the 18 year olds upset.
They're gonna they're gonna want blood,

(36:51):
but
there's a reason for it.
There's a reason for it.
Do you stand with them making sure that
under eighteens are not being
subjected to certain content?
The new DJI mini five pro leak suggests

(37:14):
it could be a perfect travel dome,
200 under 250 grams.
That's kind of a key weight here.
It should be the first sub 249
gram
249
gram drone with a one inch sensor.
Battery appears to be same as used by
the DJI Mini four and Mini three Pro.

(37:36):
So series of pictures have been released.
And,
so, anyway, those who've been waiting for the
new one, it's,
they got picks,
available for you to look at in, in
the article.
Samsung may be being beat to market on
their smart glasses.

(37:57):
HTC,
the company, which for several years brought us
beautiful phones including HTC One, One m eight,
and the One m I One One m
nine
before withdrawing
from the market
to focus on its VIVE VR headset is
now making splash in the AI Smart Glass
arena.
Samsung is trying to get their eyeglasses to

(38:19):
market as well.
So,
we're gonna see a plethora of these types
of products,
online soon.
Facebook had a bit of a a blip
today starting about 2AM
Pacific
to 08:30AM
Pacific. So many of you were probably asleep
when this happened.
I wasn't on Facebook at all. I I

(38:40):
don't get on as much as I used
to for sure.
US weather agency dangles 396,000,000
to run operations for its next
space watching fleet.
Hurricane data,
sun data
goes on and on.
And,
the contract is seeking a multi mission operational

(39:01):
service architecture support all space weather next space
weather environment observations.
So,
link will be up in the show notes
on this.
Desktops and printers in coffee shops, Starbucks
Korea tell customers
no.

(39:21):
Now I've seen all kinds of crazy things
in coffee shops, but personal desktops, printers, power
strips,
partitions
cannot be used in a store in South
Korea. Not allowed. But laptops and tablets are
welcome.
New York State has sued Zelle,
saying the Security Labs led to a billion

(39:42):
dollar in consumer fraud loss.
So they're going after Zelle for,
for lost money.
And,
many, many banks
have disconnected from Zelle
and will not reconnect with them.
Google is pledging 9,000,000,000
to expand
AI cloud infrastructure in Oklahoma.

(40:04):
So,
those of you in which area is it?
Stillwater?
Stillwater,
Oklahoma
prior
facility.
So getting a big investment there.
Again, more AI
data centers.
AI demand boost iPhone maker Foxconn second quarter

(40:25):
profit by 20 per 7%
beating,
estimates.
And, the company forecast its revenue from its
AI server business could grow over a 170%
year over year during,
moving forward.
So,
they had 1,790,000,000,000.00

(40:46):
new Taiwan dollars or 59.73
US.
Net operating profit of,
49.76.
Wow.
Operating profit was huge.
So they're making some serious money there, Foxconn.

(41:07):
Italian hone Italian
oh, excuse me.
Italian Hotels
breach in mass since June. Government confirms in
a 100,000
records allegedly after sale that apparently,
after apparent breach.
So,
more cyberattacks,

(41:27):
of course.
And,
the iPhone 17 Pro, all the rumors and
leaks so far,
I have them all linked up from, tomsguide.com.
Haven't had a tomsguide.com,
article in a long time. Long time site.
Give it some support,
when you're looking to the, the show notes

(41:47):
today. As always, do not forget about our
sponsor. We want to thank you very, very
much for your ongoing support of of GoDaddy,
keatoncentral.com/godaddy.
Of course, you can reach me on x
at geek news, and my email address is
[e-mail].
[e-mail].
It's been my pleasure to bring you the
show. Again, sorry. I'm off for the next

(42:09):
two again,
but, headed to Dallas
and,
for podcasting events
and for podcast conference,
speaking three times,
during the week.
Lots of activities going on, hobnobbing, catching up
with people, and hopefully talking to a lot
of new customers.
We'll talk about more about that when I
get back. But, everyone, thanks for being here.

(42:31):
It's good to be back in the good
old US Of A in the studio.
Everything came up and worked pretty good on
the new media show. We had a little
bit of streaming issue,
but it looks like,
everything has been online and cooking tonight. So,
again, thank you for being here. If you're
watching the show live, make sure you leave

(42:51):
a comment before you go. It's been my
pleasure to bring you the podcast. We'll see
you back
here,
a week from
Monday.
Everyone take care. So see you next time.
Bye bye.