We'll be right back.

Welcome to another episode of Cloud

Unplugged.

We have some more news, as always.

We have Anthropix call to stop and

regulate AI,

which we'll kind of come on to.

We have Instagram getting hacked by Meta's

own AI support chatbot.

We've got a little bit more on Meta

around the cloud computing aspect of them.

Are they going to do it?

Are they not going to do it?

Are they going to get into that space?

And then we've got the Monash chip,

which is basically using photonics for

transmitting data on the chip.

And then we have two random stories.

And as always,

we'll kind of save them to the end.

um but yeah the anthropic uh basically the

call to stop ai by a company that

does ai that's ipoing um what are your

thoughts there uh salmon so anthropic has

come up last just last night and said

oh we need to stop doing what we're

doing make sure we have the right

regulations

Call me a cynic, John,

but this sounds to me like the classic

pull up the ladder,

as the economics people call this.

We've made some improvements.

We've got some market share.

Let's just pull up the ladder to make

sure nobody else gets into it.

That's what it seems like to me.

I mean, it's happened before.

In the past,

telecoms companies have done this in the

past.

AT&T, one of the big ones in America.

Railway companies in the eighteen

hundreds, they did that as well.

Pharmaceuticals do it all the time.

They do it all the time.

They ask for regulation.

It sounds a little bit weird because

they're asking the world to slow down the

AI development.

And they actually also said that eighty

percent of the code is now written

by AI,

so their researchers are not really

writing the code anymore.

Eighty percent of it.

But sounds a bit fishy when you do

an initial public offering for one

trillion dollars,

and perhaps you want to make some more

noise

Maybe you stop mentioning this stuff.

I'm not really too sure.

We haven't really heard too much from

OpenAI,

the likes of OpenAI and the other

companies.

But it sounds like perhaps a little bit

more marketing for me,

from my point of view, John.

You can call me cynical.

But for sure,

we definitely need regulation on how

things are being built,

how safety needs to be taken into

consideration.

It sounds a little bit of a marketing

stunt for me, John.

I don't know what your thoughts are on

that.

Well,

as somebody that's about to invest in

them, I think they need to regulate.

I think I'm all for it.

I think you should be listening to them,

Simon.

You know, they're about to.

I don't know, actually,

because to be fair, Jack Clark,

he was a tech journalist and

moved to San Francisco.

I know a little bit about him.

He moved to San Francisco and was obsessed

with AI and the progression of AI.

I think back then,

even while DeepMind was being produced,

there was a cool...

I can't remember the guy's name that was

driving DeepMind,

but

Yeah.

Even they were saying to regulate back

then.

And he's been a bit of an advocate.

I don't think this is a new thing

for him to say.

He's said it for quite a long time,

just generally speaking,

and has been calling not necessarily for

regulation to... I mean,

this one feels a little bit fishy because

of the way it's been phrased.

Yeah.

but he was talking previously about how,

you know,

like children's toys and the safety of

them or food standards and the safety for

food that you eat and things like that.

And actually saying, you know,

at the moment there is nothing to stop

anybody doing anything.

So the mythos model is,

They obviously didn't have to restrict it.

There's no regulation to stop it.

They could have just let it fly.

As any company that builds any model

could,

there's nothing stopping you today

building some insane model that can,

I don't know,

specialize in biological warfare or

something, right?

And you could train that today and launch

it,

and there's nothing preventing you doing

that.

And then the ramifications of that is

obviously like unbound.

So I know he does believe in those

things and he think he does want checks

and balance.

He doesn't want to slow it down,

but he wants the checks and balances that

proves that it's safe,

that proves that this model is safe.

It can't be used for malintent or maybe

there's exposure internally and tests and

things by government or something to

assess risk.

I'm not really sure, but yeah.

But then this, though,

is a bit different because what he said

here, though,

is he's got a different thread where he's

then said it's about self-improvement.

And he's like,

once it starts self-improving,

then basically it can just make itself get

better.

And it's like...

improvement.

What do they define this self improvement?

How do you define self improvement?

I think it's where the model can become

more effective,

but it doesn't require engineers to do it.

It can work it out itself and then

fix almost itself to improve the training

of itself to get more efficient and

effective.

then improve the code around itself as

well and the exposure of that capability

and i think then it's almost like

basically autonomous which then means

obviously if you think at the moment the

time it takes engineers to figure out how

to train it and look at the data

and do the parameters that's human so it's

it's got a it's slower obviously because

whereas this becomes you know the growth

and the

of the model and the improvement of the

model will be like rapid because it's

basically got twenty four hours a day,

seven days a week,

nonstop could be improving yourself.

I think it is the thing.

Yeah.

Because that's based on a criteria that's

been defined already by somebody that says

this is your gain or loss that you

need to get.

And based on that, it self improves.

Again, call me cynical, John, if you like.

I always don't really understand what the

self improvement means in its sense.

just producing more lines of code.

Is that self-improvement?

Yes.

You know, in neural networks,

you have the loss function that you need

to minimize in order to get the right

answer.

But does that going to result in business

value that's being added?

I'm not sure.

I don't know what the self-improvement

actually means.

I have no idea what that means, John.

So perhaps you have... Yeah,

I guess because we're doing some... Well,

I know they've got...

telemetry.

So they've got a lot of open telemetry

internally on how effective their model is

being used with anonymized data of

customers.

So then they can kind of see,

and I think there was talk about exposing

that data economically to see if that's

correlated with actual layoffs of

companies to then see if really it's AI

that's causing disruption to basically

people laying people off or are they just

using it as an excuse because they're

bloated and there's no evidence to suggest

that it is really AI.

And so there's that.

So I imagine the same data.

Obviously,

they'll have the data of how good their

models really have been.

But then if it can write code and

improve itself,

the model just gets better and better and

better and better.

Fair enough.

I presume and then can write its own

code because it's just basically

Yeah, like us,

we just get better and better and better

every day.

Better predicting the next token.

That's what it's doing.

So my question is, yes,

all these companies for a long time,

even before any of this Gen AI stuff

was a thing,

people were asking for more regulation and

safer building of AI.

But nothing really has happened, right?

Yes, we have research groups.

who's actually going to start implementing

these rules right so if you let's say

if you go formula one fia their their

governing body that has rules that the

companies have to follow in terms of how

much they can spend on development and the

regulations that they provide they provide

strict regulations that this we need to

build against at the car the downforce

levels the engine power who's going to

regulate

AI, who's coming up with all these rules?

Is it representative of all these

companies?

Is it going to be like an open

source board, kind of like CNCF,

Cloud Nading Computing Foundation?

How it's going to work?

I don't really understand.

Is the government going to be involved?

Are these companies that are building the

models they themselves are going to

regulate?

So how is that going to work?

I don't really understand that part

either.

more questions than answers john yeah i

don't know because in uh and the thing

when i was like listening um to what

he was saying recently i don't know how

recent maybe a week ago or maybe a

couple of days ago but um the a

bit like he was basically saying like a

nuclear power plant

has the ability to create nuclear bombs.

But you don't stop the building of the

nuclear power plant.

You regulate those things so that it's

less harmful.

But it doesn't mean that you shouldn't

build them.

Well, it did mean that historically,

actually.

So it actually slowed them down because of

the risks of what happened.

And I guess like anything who comes up

with the standards,

social media currently,

is it bad for children?

How do we really know what impact is

it really having?

Should that have been regulated before?

Should there have been a better

understanding of it?

And I suppose on a small scale,

that's niched into obviously children and

maybe then becoming adults and what impact

that's having.

But that's more social.

This is more economical.

It could be also defense, cyber crimes.

It could be the production of all kinds

of things.

If it's weapons, bombs, new viruses,

all kinds of stuff that you could,

in theory, be an expert on,

which you couldn't have obviously managed.

So I do...

kind of understand,

should you be able to get those types

of answers if you aren't a scientist?

Should you be allowed to go work with

a model that then can teach you how

to build biological weapons if you have

nothing to do with that field?

Is that safe and good?

And is there a general use case for

people learning about that things just out

of self interest?

And I guess that's probably who decides

what information is exposed to who and

what's exposed to companies that may be

it should be exposed to because that's the

industry they're in, maybe.

I don't know.

I think that's the... Yeah.

No,

it's interesting that Anthropic came out

and said, like,

we need to slow down the development of

AI.

But after they got a fairly decent market

chunk... I absolutely agree with you,

though, by the way, just to clarify.

I believe this is to do with the

IPO because all the things I've just said

are not what he's just said.

These are the things I know they've said

before.

But this time they're like, oh my God,

look how good our model is.

It's already right in eight percent of our

code.

Soon it's going to be a hundred percent.

I can't believe it.

Wow.

It's so self-improving.

It's like the best model out there.

We should regulate it.

By the way, we're IPO-ing,

if anyone's interested.

Yeah, so yeah, that's totally different.

Yeah,

it's like being ahead on one nil in

football and the sixty-seventh minute.

You're like, oh, ref,

just call it off now.

The game's over.

It's done.

It's interesting about this IPO stuff as

well, John.

I don't know if you've seen

recently uh nasdaq you know the the index

providers like that nasdaq and footsie

russell they've kind of changed the rules

around um once when the company's ipo they

have what they call a seasoning period

where between like three months to a year

before you can take any of these your

ipos the you know the shares and put

them in indexes right so three months to

a year

To entice these big tech companies like

SpaceX, like Anthropic to IPO,

the Nasdaq and FTSE Russell have dropped

down this seasoning period from like

months, literally down to days,

about five days or ten days.

right and also uh things like like nasdaq

have a minimum uh free float rule of

like ten percent even that got scrapped

and also these companies uh will if they

do float they'll get a three-time

multiplier so one of their shares is about

three so something really interesting is

happening in economics world to entice

these companies and perhaps you know they

had a

discussion but I think it's still

developing this news around changes in

market the way markets work just to entice

these companies to IPO because there's a

lot of hype around it I think that's

the only thing I can put to it

there's literally a lot of hype around it

so anything these companies say people

like Anthropic or SpaceX or OpenAI it just

becomes like I have to look at it

is this yet another marketing

Really?

So is that what we're doing here?

Well, I don't know.

This is absolutely marketing, isn't it?

Let's be honest.

Even like Trump, I think, David Sachs guy,

they said the same thing.

Basically, it's just a genderistic.

They're just trying to capture their own

market and then obviously cause it to

squeeze so that they can obviously

capitalize.

But I don't know whether it's really that.

I think it felt like

they've done a bit of a clever PR

stunt where they've promoted through

something negative.

Do you know what I mean?

Like, look how amazing we really are.

We should do something about this, guys,

because at this rate, we're so amazing.

God knows what's going to happen.

Do you know what I mean?

That's kind of like the thread, isn't it?

If we continue being this amazing...

Who knows?

Who knows what damage you might do?

Somebody stop me.

But yeah.

Yeah, fair enough.

But at the same time,

all the companies are cutting down on

their AI spend.

Uber have already,

their CEO came out and they said, well,

we talked about this last week,

that they spent all their

their ai budget in april and they're

limiting it to fifteen hundred dollars per

developer person fifteen to a hundred

dollars worth of tokens and the same thing

is happening in microsoft so i think uh

i think anthropic is like well before it's

too late let's just ipo and make the

money i don't know i'll probably gonna get

cancelled but i'll probably buy some

shares in anthropic myself no because we

were looking i mean to be fair like

this isn't just anthropic but

We were looking internally recently on

like, okay,

we're going to have an internal project.

We're very small.

We're not like a big company.

We're not an Uber, right?

And we want people to use AI.

We want to create like some agents could

be using whatever.

It could be using Agent Core in Amazon.

It could be using the agentic service in

Azure, whatever.

It doesn't really matter what it is.

But in the cloud providers, we're like,

okay, well, how do we stop it?

How do we stop an accident where somebody

does writes the agent and it basically

could almost like go through every single

piece of data in an S three bucket

indefinitely and kind of continue looping.

Right.

So we have to be careful.

So let's put a cap on it.

You can't put a cap on it.

So basically.

It's like several hours before the billing

data will arrive anyway.

And obviously you're talking about

instantaneous tokens and all you can get

is like the token metric and then you'd

have to work it out yourself.

So you'd have to build something to

somehow figure out whether you are or

aren't hemorrhaging money inside of the

cloud by accident with something using

agents and LLMs and things like that.

And that's not even the developer side on

top of it.

Obviously you've got developer usage.

It feels like the cost to value ratio

without the right constraints means you

could quite easily

just burn through something.

You could spend hundreds of thousands of

pounds for something that didn't need to

cost you hundreds and thousands of pounds,

if you see what I mean.

Let me use all my tokens just to

center a div.

Let's do that.

Yeah, exactly.

And then it actually costs you way more

money than had you just built it the

old way.

But yeah, I think it's a bit strange.

But I don't know.

I definitely don't think the

The hype's over.

I think we've only just started,

to be honest.

Oh, yeah.

China's done some big investment as well

in their AI startups and data centers.

And so the deep seek, I think,

got huge investment.

I think they've just done a huge billions

in deep seek.

So I kind of yeah.

It's just poor little us, poor little us,

poor little island in the UK that's got

nothing.

We've got no data centers,

got no real models.

You know, what have we got over here?

BP?

BP, yeah.

Yeah.

Octopus Energy.

Where's our models?

Yeah, it's not looking good.

But what is good,

and let me just tell you something that

is pretty good,

is if you get locked out of your

Instagram account, you can now,

using Meta's new feature, get in it.

No problemo.

You get locked out and you need to

get in,

it's letting you in without any problem.

Not just your account, anybody's account,

for that matter.

So, yeah, if I get locked out,

I can ask you, hey, Simon, I'm...

I'm locked up my Instagram account.

Can you get in for me?

And you'd be like, yeah, no worries, John,

I can do that for you.

So this is the news that Meta released

a new AI powered chat support agent,

which you can talk to to reset your

password and log in if you get locked

out.

And what people found was you can sweet

talk this agent.

reset your reset the password for some

somebody else's account so what people did

was they went on connected started

chatting with this ai powered agent and

said yeah my account has been hacked and

they just picked a random account uh i

think they picked one of the obama's era

white house account from instagram it's in

an active for for a long time but

it's one of those accounts

And they said, well, I've been locked out.

Can you just reset my password?

Here is my email address.

Provided a completely new email address.

And that email address received a token

for logging in.

And you put the passcode in,

and it gave them an access.

And it didn't impact any accounts that

have two FAs,

two-factor authentication enabled,

but the chatbot was able to bypass all

of the setup and configuration that you

used to do in the past where only

the right people can access it and confirm

it.

People were able to sweet-talk this agent

and get into accounts very easily.

So that's what happened last week.

That's crazy.

That's crazy.

should I say the word dumb?

If you are going to do a chatbot,

you wouldn't think it would be over,

you'd think it would know the identity of

the person talking to it to know that

what they've asked for isn't pertaining to

that.

It doesn't seem that complicated, but

yeah or just stop the chat bar from

doing any identity related stuff yeah you

know this is this is straight off the

instagram meta sorry metagram meta

announced they laid off a lot of employees

so of course they're looking into

improvements efficiencies using ai i think

this is another in another case where a

feature was released way too quickly

perhaps

I feel like maybe it wasn't tested that

well.

Because you can put guardrails in models

and tell it, oh, this is your boundary.

You cannot go out of this.

You cannot do.

You build into that.

You cannot do any of the identity-related

stuff.

Yeah.

If you do want to confirm it.

And also there were instances where people

were saying they managed to create videos,

video selfies of people using AI tooling

to fool this AI instant,

AI chat support model to say that they

are the actual person just using their

photo.

So yeah,

I think it's probably just way too...

They just released the word too quickly

and caused a bit of an issue.

But many of the people,

because they had multi-factor

authentication enabled,

that kind of worked in their favor.

Security, as always,

has to be done in layers.

But this just seemed, as you say, John,

very dumb in the way that happened.

And it caused a bit of a nightmare.

But this is the thing with AI.

nowadays, John,

because there's so much news.

People forget really quickly.

This happened a few days ago,

like three to four days ago.

It didn't happen that long ago.

I didn't even see this one,

which is what's quite hard.

It's just almost like so much going on.

You kind of can't even really keep track.

I didn't hear of this one until you

told me about it.

Yeah, I mean, fair play to Meta.

They fixed the issue.

This happened, I think,

end of May and they fixed by the

second of June.

It was fixed in a couple of days

because, yeah, of course,

you can stop a chat from doing those

kind of things.

But where's the testing?

Why was this not tested?

Well, they did.

That was the testing.

production.

Yeah,

there was a test and they got in.

They were like, it's the Obama test.

That's what it's got.

If you either can get in the White

House or hack Obama's account or whatever,

that's the true Obama test.

Or even trick people into thinking you're

Obama.

It's the other one with AI.

And that's it.

So basically what happens is attackers

found they could steal the Instagram

accounts by chatting to Meta's own AI

support bot and asking it to hand over

access.

So you don't need any password.

The Obama White House account went.

A Space Force official account went.

A security researcher account went.

The bot was just basically trying to be

helpful.

Turns out being helpful...

And being able to verify who you're

talking to are two completely different

product requirements.

And what Meta did was ship only one.

Be helpful.

And that's what happened.

What actually happened to those accounts,

though?

What did they do to those accounts?

Did they, like...

Yeah,

because some of these accounts were

apparently were created like a long time

ago.

So they have short handles.

I think people try to sell it.

Perhaps.

I'm not sure what.

I have followed on to onwards.

What happened to it?

People can try and sell these accounts,

right?

Because the handle has followers.

It's a good handle.

And it's a good handle.

You were also mentioning about the data

center cloud computing matter entering

cloud computing before to me.

And I know there's been talk rumors of

this before.

I think they've made suggestions of this

before a while back.

I can't remember when,

because obviously they've got all their

compute there and then maybe they could

like do an Amazon where it's like, oh,

we could turn into like a cloud provider

type thing.

Do you know more about this?

Are they going to do it?

Yeah, I mean,

at Meta's annual shareholder meeting on

May, May, May, May, May, May, May, May,

May, May, May, May, May, May, May, May,

May, May, May, May, May, May, May, May,

So, you know,

they've not rented out any of the

capacity.

Don't forget Meta, you know,

they've built a large infrastructure for

themselves.

They're on data centers so they can host

tools like Facebook,

like WhatsApp and Instagram with millions

and millions of users.

And also they have their own open source

model stack for Lama, you know,

the open source models that last got

released in twenty twenty four.

And I think they're reworking on a bunch

of stuff.

And, you know, they they're

they they've never rented stuff out before

but now the capacities are running out

elsewhere and the calculation is changing

because you can earn a lot more money

and uh meta has raised its twenty twenty

six ai capital up to like hundred twenty

five billion dollars uh and what they're

trying to do is also uh secure a

kind of like six gigawatts of nuclear

energy to powers data center so

And in the current cloud market,

you really have three big players.

AWS, about thirty percent of the share.

Azure, about twenty five percent of share.

Google, around ten percent of share.

And, you know,

Meta's thinking perhaps we can also become

one of the big fours, you know,

instead of.

So that's what that's what they're doing.

It's an interesting move.

I don't know if it will happen, but.

they have a lot of CapEx that they

can use to build this and perhaps maybe

even undercut Azure and AWS to get some

customers yeah I don't know because I know

that like we mentioned last time Google

got a load of investment for obviously

their CapEx spend so they're like

de-risking and I think

Obviously,

even with some of the IPO-ing and other

things, obviously,

that's other ways to kind of offset your

own costs even.

So I think there's like a little bit

of a, I don't know,

are people a hundred percent convinced

that they're prepared to risk their own

financials and money on the capex spend of

AI compute for the market?

And it feels like they're de-risking

you know,

the gamble by using other people's money,

VC's money and getting investment in and

things like that,

which would beg the question of, you know,

I don't know what their services even

would be outside of just native compute,

you know, how they've automated, because,

you know,

you've got to make it a good experience

as well.

You can't just say you've got like racks

of compute, you know,

some could be easy to consume.

But yeah, I don't know.

I don't know.

It seems a bit,

suspect.

to me that I feel like people are

not sure.

I think a lot of the companies are

just kind of not too sure.

But it's another revenue stream, John,

because at the moment that's not really a

revenue stream because they're using it.

They perhaps realize that they have

capacity.

That's kind of like AWS, right?

They realize they have capacity laying

around.

Maybe they can sell it.

And because they went heavily in twenty,

twenty three, twenty,

twenty four into the open Lama models,

But they're not releasing anything.

They haven't released anything for the

last, like, eighteen months.

So I'm not sure what's happening

internally.

Perhaps they're like, well,

let's just sell this compute.

I think it'll be good,

but I'm not sure if Meta will have

their own AI agents that you can get

into other people's account.

You'll be able to just host really easily

on other people's compute.

You just chat to the bot and just

be like, put me into Salmon's tenant.

Kubernetes cluster.

Yeah, exactly.

Get it in his Kubernetes cluster.

Yeah.

So there's another story I kind of moving

on,

which is I don't have a huge amount

because it's like a little bit of

innovation happening.

But basically there's been the Monash chip

and it's a way of using photonics,

but basically creating light essentially

to basically transmit data,

but then kind of bending it and curving

the light and then transmitting it through

to the receiving end.

basically an electrical signal,

all in one device, essentially.

So meaning that they can obviously

transmit.

One,

it's very small because it's atomic size

level.

So it's like a special type of light.

And it's like atoms, like very,

very small atomic size nanostructures.

And therefore they can obviously like

transmit the data in much smaller amounts.

of chipset and obviously a high volume of

data using photonics and basically

photonic circuits which are like less

dense than like optical circuits and

things like that but that's all i know

a little bit of innovation from a

university.

I think it was Stanford University

potentially.

Yeah.

I might have got that wrong.

But yeah, which is kind of cool.

A little bit of innovation that isn't AI.

What's the main benefit around it?

Does it use less energy?

Or is that the main thing?

Of course, it's faster because, yes,

it's light.

But is there an additional benefit?

Perhaps it doesn't use as much energy as

silicon chips.

well the obviously it doesn't heat it's

not using copper to transmit it's using

light so if you're using light everywhere

then it doesn't heat up so it's obviously

room temperature you don't need to cool it

as much um so there'll be benefit there

but all the peripheral things who knows on

like how it would wire up into actual

compute chips in the end and what the

chip even is it's probably not a very

you know the transistor on it is probably

quite meaningless

But yeah.

But that was like another little bit of

thing.

So things kind of going on.

Always innovation happening,

which is good.

It's not all AI innovation, Salman,

all right?

No, it's good.

It's good.

Yeah,

I want some quantum and I want some

light now, John.

You need some light.

You need some quantum.

And maybe there's a thing called

Kubernetes.

We could maybe talk about the...

The K eight as people abbreviate to first

we got to learn K eight and then

we can talk about it.

I don't learn K eight nowadays.

I don't learn K eight.

Yeah, that's the right way forward.

I think that's fair enough.

Yeah, and you just clawed it.

Just learn clawed A-Eights.

But anyway, yeah.

So, and then some random stories.

So,

do you want to talk about your random

stories of the...

Yeah, my random story is related to AI.

And this is the season of graduations.

People are getting graduated around this

time.

Especially in America,

you have commencement ceremonies in USA in

like May.

And graduating students are listening

usually to a speaker who comes in and

they will give a speech and talk.

gave some words of encouragement on how

they can go into the world and make

a difference.

But what people noticed that every time

somebody mentioned AI,

the students started booing whoever the

speaker was.

And it was a bit funny because a

former Google CEO, Eric Schmidt,

did it as well.

He started talking about how AI is shaping

the future and how people need to look

into it.

And people started booing it.

There's another commencement ceremony

where you had an automated AI announcer

that basically mispronounced the names and

skipped a bunch of students' names as well

because they're like, oh, it's AI,

so we should use that.

That didn't go down too well.

But the point here is that unemployment

amongst recent graduates between ages of

twenty two and twenty seven has gone up

to like five point six.

And entry-level job competition has

increased by like, twenty percent.

And I think people,

students are perhaps a little bit weary

when people start mentioning AI because

we've seen, we've talked about this stuff.

But yeah,

and one of the commencement speeches when

people are booing, the speaker just said,

deal with it.

I mean,

that's not the right way of doing this

stuff.

But it's interesting to see what's

happening.

You know, this is a bit of a...

uh random news which is not so random

in in our current climate that uh the

people are booing the message of ai i

don't know why they're booing like what

they're doing about just become an

influencer that's like what is it to boo

you can play games

Make money from people watching you.

Just become an influencer.

That's it.

That's all you need to be focusing on.

Start a podcast.

That'll work too.

Film every detail of your life.

Even influencers, they're becoming AI now.

People making AI influencers.

What have you got left, John?

You were asking this question.

People follow AI influencers though.

The prime job...

you could become a dentist or a vet

or, you know, you could do, you could,

you could find a vocation,

maybe work with other people,

find out what you're really good at,

get social skills, get friends,

or just don't just become an influencer.

Focus on like recording yourself eating

and every thought you just record,

you put it out there.

You,

you basically film all your thoughts and

that's how you should be.

That's where you should be spending your

time.

That's, that's, that's the future.

Thanks for the advice, John.

I like to give good advice.

My story, which is a quick one.

If you've got a phone and you've got

a bit of Bluetooth on there and you're

deciding what your device might be called

and you're enabling it,

maybe for your headphones or whatever,

don't call it bomb when you're on a

flight because

If they see it,

that flight's turning around.

And this happened from going to the United

States.

So it's going to Mallorca,

Palmer in Mallorca.

And basically, they U-turned.

So they put an announcement saying,

whoever's Bluetooth device that's called

BOM, can you turn it off?

And they didn't turn it off.

And then so they basically turned the

entire flight round and went all the way

back.

the united states um because they were

worried about someone's uh phone uh maybe

it was like real or something i don't

know um so yeah so you got to

be careful with what you're labeling your

phones you can't just call it whatever you

want um so there you go a little

lesson for people and on that bombshell

it's time to end the episode i like

what you did there like what you did

that was good that's smooth

Smooth operator.

Cool.

But yeah, that is time for the episode,

isn't it?

Actually, to be fair,

that is all the news.

So yeah,

tune in next week where we'll have more

news happening as always.

And we should keep you posted on what's

going on.

Cheers.

Cool.

Bye-bye.